(Resolution will improve progressively)

dv said:

(Resolution will improve progressively)

It’s Edward of Woodstock, known as the Black Prince, the eldest son of King Edward III and Philippa of Hainault, and the father of King Richard II of England. He was the first Duke of Cornwall (from 1337), the Prince of Wales (from 1343) and the Prince of Aquitaine (1362–72).

dv said:

(Resolution will improve progressively)

My eyes hurt.

Bubblecar said:

dv said:

(Resolution will improve progressively)

It’s Edward of Woodstock, known as the Black Prince, the eldest son of King Edward III and Philippa of Hainault, and the father of King Richard II of England. He was the first Duke of Cornwall (from 1337), the Prince of Wales (from 1343) and the Prince of Aquitaine (1362–72).

8-O

Bubblecar said:

It’s Edward of Woodstock, known as the Black Prince, the eldest son of King Edward III and Philippa of Hainault, and the father of King Richard II of England. He was the first Duke of Cornwall (from 1337), the Prince of Wales (from 1343) and the Prince of Aquitaine (1362–72).

But no that’s not who it is. Pretty close though, in that this is a picture of a human being.

8 × 8

I’m assuming this is a profile view.

Prince Phil

Well, this is exciting.

kii said:

Well, this is exciting.

IKR

I’ll start again in the moz when people are here

Hitler.

dv said:

kii said:

Well, this is exciting.

IKR

I’ll start again in the moz when people are here

I am in the moz. I am a person.

kii said:

dv said:

kii said:

Well, this is exciting.

IKR

I’ll start again in the moz when people are here

I am in the moz. I am a person.

He wants whole crowds of people bickering about it.

“Hitler!”

“No, it’s not Hitler. It’s that actor who played Hitler, whatsisname.”

“Rod Steiger.”

“Rod Steiger never played Hitler.”

“It’s Hitler.”

“Looks more like Elvis.”

“No it’s whatsisname, Rod Steiger.”

“Rod Steiger never played Elvis.”

“It’s fucking Hitler I tell you.”

etc etc

Well it is a kind of game and I’m sure it will be more enjoyable with more people playing. I doubt there’d be any bickering, people here are so nice.

Konstantin Tsiolkovsky.

dv said:

(Resolution will improve progressively)

The resolution improvement does not look very progressive to me.

Richard Simmons…

dv said:

At first I thought I’d seen this face before, now I’m not so sure.

A woman, with ponytail or similar hairstyle. 3/4 view facing left. Has unusually bright cheeks and a narrow chin. Doesn’t wear glasses.

dv said:

At first I thought I’d seen this face before, now I’m not so sure.

A woman, with ponytail or similar hairstyle. 3/4 view facing left. Has unusually bright cheeks and a narrow chin. Doesn’t wear glasses.

Also, it’s a surprisingly narrow face in general, which suggests that it may (or may not) be a cartoon or caricature.

The following image is the most famous pixellated face. I would be surprised if you can’t recognise it.

mollwollfumble said:

The following image is the most famous pixellated face. I would be surprised if you can’t recognise it.

Be surprised.

But I do have a guess, which I will keep to my self.

In spite of not recognising it, it is always sort of surprising how you see more “detail” when you squint at a low resolution image.

The Rev Dodgson said:

mollwollfumble said:

The following image is the most famous pixellated face. I would be surprised if you can’t recognise it.

Be surprised.

But I do have a guess, which I will keep to my self.

In spite of not recognising it, it is always sort of surprising how you see more “detail” when you squint at a low resolution image.

I’m no good at this sort of thing. The only face I can see there is Ned Kelly wearing his armour.

The Rev Dodgson said:

mollwollfumble said:

The following image is the most famous pixellated face. I would be surprised if you can’t recognise it.

Be surprised.

But I do have a guess, which I will keep to my self.

In spite of not recognising it, it is always sort of surprising how you see more “detail” when you squint at a low resolution image.

Squinting gets me to LG.

dv said:

http://www.dazvoz.com/bw008.jpg

At first I thought I’d seen this face before, now I’m not so sure.

A woman, with ponytail or similar hairstyle. 3/4 view facing left. Has unusually bright cheeks and a narrow chin. Doesn’t wear glasses.

Also, it’s a surprisingly narrow face in general, which suggests that it may (or may not) be a cartoon or caricature.

Also, light brown hair.

If I’m wrong and it’s a man, then he has no facial hair, is starting to go bald at the front and is wearing a wig-like hairstyle

mollwollfumble said:

The following image is the most famous pixellated face. I would be surprised if you can’t recognise it.

The Mona Lisa

dv said:

mollwollfumble said:

The following image is the most famous pixellated face. I would be surprised if you can’t recognise it.

The Mona Lisa

If it is I claim first, I posted LG to keep the game going, Lisa Gherardini AKA The Mona Lisa.

11 × 11

dv said:

dv said:

http://www.dazvoz.com/bw008.jpg

Wild guess. A young David Letterman? I can’t tell at this resolution.

Tony Abbott (squinting, I see a male, big ears and a clown nose)

dv said:

mollwollfumble said:

The following image is the most famous pixellated face. I would be surprised if you can’t recognise it.

The Mona Lisa

I thought of that too but only from the text. Either Mona or Princess Di.

dv said:

11 × 11

dv said:

charles.

Bogsnorkler said:

dv said:

11 × 11

dv said:

charles.

or someone else.

Bogsnorkler said:

Bogsnorkler said:

dv said:

11 × 11

charles.

or someone else.

So my 3 guesses are 1) Ned Kelly. 2) Princess Di. 3) Mona Lisa.

does it get better if you move right back?

I’d say it’s a criminal or a dentist.

dv said:

11 × 11

dv said:

At this resolution I’m more sure that I’ve never seen this image before.
It looks even more like a cartoon or caricature. Very narrow face.
I think it’s got to be a cartoon or caricature because the nose is set really high on the face and the mouth really low.
3/4 face looking left and neither up nor down.
Narrow face with prominent lowset ears.
No facial hair and tending to balding at the front.
Instead of a ponytail, the hairstyle now looks a bit more like a topknot. Not quite like a mohican.
Looks more like a man than a woman.
Light brown or mid brown hair.

… or …

Perhaps what I’ve been taking for the mouth isn’t a mouth at all, but instead a throat, in which case what I’ve been taking for the narrow chin is actually the vee-shape of a collar.
Putting my hand over the bottom two pixels, the face now appears to be a beadless man with wide chin.
Protruding ears now look even more protruding.
Then what I’ve been thinking of as a ponytail/topknot could be actually background scenery, and the man has a crewcut.
With the rethink, now the man is looking more straight on (half way between 3/4 view and straight on) and slightly upwards.
There’s no clear separation between the bottom of the nose and the top of the mouth – so there could be a Hitler-like moustache, or it could just be the shadow of the nose.

Ronnie Barker?

AwesomeO said:

dv said:

mollwollfumble said:

The following image is the most famous pixellated face. I would be surprised if you can’t recognise it.

http://i120.photobucket.com/albums/o162/DavidPaterson/SSSF/lattice_zpsbtoithgo.jpg

The Mona Lisa

If it is I claim first, I posted LG to keep the game going, Lisa Gherardini AKA The Mona Lisa.

Yep.

Resolution will get finer as we progress. it is a scene.

Pixellated Face Disease (or PFD) is a strange and disturbing phenomenon whereby an individual’s face becomes blurred and blocky for seemingly no apparent reason. Any individual, no matter how law-abiding or innocent can contract the disease whilst going about their everyday life.

However, the single most disturbing aspect of the condition is that , much like ghosts, the effects can only be seen when the infected individual is viewed on film or photographic format. Thus, one will never know if one has the condition until they appear on World’s Wildest Police Chases or some such.

AwesomeO said:

dv said:

The Mona Lisa

If it is I claim first, I posted LG to keep the game going, Lisa Gherardini AKA The Mona Lisa.

Yep.

According to a test on pixellated faces that appeared decades ago in Scientific American, the Mona Lisa was the face they tested that was most recognisable at low resolution. The second most recognisable was George Washington.

Many symptoms have been witnessed in conjunction with PFD. Following is a selection of the most common:

Monsterism, also often referred to as Zombification. A process whereby an individual takes on the characteristics of a monster such as a grizzly bear with a double-edged sword, most notably a deep, low droning voice. The person in question also begins to move much more deliberately, almost as if they were in slow motion.

Pixellated genital disease. Often the individual with PGD will display this condition as well. Just as grave as PFD, it can be witnessed when the individual takes their pants off on film. Since the early 1990’s, this condition has been seen increasingly in streakers.

Blurred registration plateism. The affected individual will usually contaminate any vehicle which they drive with a similar affliction. This mainly affects the number plate, which takes on a similar pixellated form to the individual’s face.

Full Body Pixellation. In some rare cases, individuals will undergo this most extreme pixellation. Sufferers have been known to take up careers as giant lego men and movie monsters without the need for costumes.

Blurred Advertisement. It is also known for the disease to spread to people’s property, from t-shirts to bottles of water. All advertisements of the item will become blurry and the individual is free from the horrible disease.

mollwollfumble said:

AwesomeO said:

dv said:

The Mona Lisa

If it is I claim first, I posted LG to keep the game going, Lisa Gherardini AKA The Mona Lisa.

Yep.

According to a test on pixellated faces that appeared decades ago in Scientific American, the Mona Lisa was the face they tested that was most recognisable at low resolution. The second most recognisable was George Washington.

The colours gave it away.

That won’t help you in my example

Ian said:

Pixellated Face Disease (or PFD) is a strange and disturbing phenomenon whereby an individual’s face becomes blurred and blocky for seemingly no apparent reason. Any individual, no matter how law-abiding or innocent can contract the disease whilst going about their everyday life.

However, the single most disturbing aspect of the condition is that , much like ghosts, the effects can only be seen when the infected individual is viewed on film or photographic format. Thus, one will never know if one has the condition until they appear on World’s Wildest Police Chases or some such.

Is that a quote from “You’re skitting me”?

Bogsnorkler said:

Resolution will get finer as we progress. it is a scene.

A charcoal sketch of the death scene from la Boheme

There is currently no known cure for the Pixellated Face Disease. Scientists are baffled by the fact the sufferer will display almost no symptoms until caught on film. The individual will go through life looking like any other ordinary person criminal. Thus, the current advice for those that fear they have the disease is to spend their lives avoiding cameras.

Whilst this is relatively easy for those living in poorer areas such as Rockingham which the camera phone has yet to permeate, it is very hard for a PFD sufferer to go unnoticed in places such as Soho or Amsterdam, where large amounts of secret cameras have been set up to make ‘home-movies’.

mollwollfumble said:

dv said:

11 × 11

dv said:

At this resolution I’m more sure that I’ve never seen this image before.
It looks even more like a cartoon or caricature. Very narrow face.
I think it’s got to be a cartoon or caricature because the nose is set really high on the face and the mouth really low.
3/4 face looking left and neither up nor down.
Narrow face with prominent lowset ears.
No facial hair and tending to balding at the front.
Instead of a ponytail, the hairstyle now looks a bit more like a topknot. Not quite like a mohican.
Looks more like a man than a woman.
Light brown or mid brown hair.

… or …

Perhaps what I’ve been taking for the mouth isn’t a mouth at all, but instead a throat, in which case what I’ve been taking for the narrow chin is actually the vee-shape of a collar.
Putting my hand over the bottom two pixels, the face now appears to be a beadless man with wide chin.
Protruding ears now look even more protruding.
Then what I’ve been thinking of as a ponytail/topknot could be actually background scenery, and the man has a crewcut.
With the rethink, now the man is looking more straight on (half way between 3/4 view and straight on) and slightly upwards.
There’s no clear separation between the bottom of the nose and the top of the mouth – so there could be a Hitler-like moustache, or it could just be the shadow of the nose.

Ronnie Barker?

It is not Barker.

I’ll tell you for free that this is not a cartoon. This is an ordinary photograph.

Ian said:

There is currently no known cure for the Pixellated Face Disease. Scientists are baffled by the fact the sufferer will display almost no symptoms until caught on film. The individual will go through life looking like any other ordinary person criminal. Thus, the current advice for those that fear they have the disease is to spend their lives avoiding cameras.

Whilst this is relatively easy for those living in poorer areas such as Rockingham which the camera phone has yet to permeate, it is very hard for a PFD sufferer to go unnoticed in places such as Soho or Amsterdam, where large amounts of secret cameras have been set up to make ‘home-movies’.

Same thing happens with some military members, no symptoms until caught on film then they display strips of black nasty across the eyes.

Walking into the room from the opposite doorway seven metres away, the 11*11 face suddenly seemed startlingly familiar, an impression that vanished in a split second.

Tamb said:

Bogsnorkler said:

Bogsnorkler said:

charles.

or someone else.

So my 3 guesses are 1) Ned Kelly. 2) Princess Di. 3) Mona Lisa.

It is not Charles, Ned, Di or Mona

Bert

Nixon

Maxwell Smart.

dv said:

mollwollfumble said:

dv said:

11 × 11

At this resolution I’m more sure that I’ve never seen this image before.
It looks even more like a cartoon or caricature. Very narrow face.
I think it’s got to be a cartoon or caricature because the nose is set really high on the face and the mouth really low.
3/4 face looking left and neither up nor down.
Narrow face with prominent lowset ears.
No facial hair and tending to balding at the front.
Instead of a ponytail, the hairstyle now looks a bit more like a topknot. Not quite like a mohican.
Looks more like a man than a woman.
Light brown or mid brown hair.

… or …

Perhaps what I’ve been taking for the mouth isn’t a mouth at all, but instead a throat, in which case what I’ve been taking for the narrow chin is actually the vee-shape of a collar.
Putting my hand over the bottom two pixels, the face now appears to be a beadless man with wide chin.
Protruding ears now look even more protruding.
Then what I’ve been thinking of as a ponytail/topknot could be actually background scenery, and the man has a crewcut.
With the rethink, now the man is looking more straight on (half way between 3/4 view and straight on) and slightly upwards.
There’s no clear separation between the bottom of the nose and the top of the mouth – so there could be a Hitler-like moustache, or it could just be the shadow of the nose.

Ronnie Barker?

It is not Barker.

I’ll tell you for free that this is not a cartoon. This is an ordinary photograph.

Fucking ordinary.

Art Garfunkel.

Einstein…

dv said:

Tamb said:

Bogsnorkler said:

or someone else.

So my 3 guesses are 1) Ned Kelly. 2) Princess Di. 3) Mona Lisa.

It is not Charles, Ned, Di or Mona

Not Charles?

I was pretty sure it was going to be either Babbage or Darwin, or some clown in between.

Her Majesty the Queen.

paul ryan.

Scott Pruitt about to be hit with a baseball bat

Are we there yet?

dv said:

11 × 11
I’ll tell you for free that this is not a cartoon. This is an ordinary photograph.

Cheating. This is the same 11*11 image when Irfanview views it full screen.

Well at least we know now that it’s not a woman.

Hair dark brown.

It might be ”Pablo Diego José Francisco de Paula Juan Nepomuceno María de los Remedios Cipriano de la Santísima Trinidad Ruiz y Picasso”

or it could be “Adolph Blaine Charles David Earl Frederick Gerald Hubert Irvim John Kenneth Loyd Martin Nero Oliver Paul Quincy Randolph Sherman Thomas Uncas Victor Willian Xerxes Yancy Zeus Wolfeschlegelsteinhausenbergerdorffvoralternwarengewissenhaftschafers wesenchafewarenwholgepflegeundsorgfaltigkeitbeschutzenvonangereifen duchihrraubgiriigfeindewelchevorralternzwolftausendjahresvorandieer scheinenbanderersteerdeemmeshedrraumschiffgebrauchlichtalsseinu rsprungvonkraftgestartseinlangefahrthinzwischensternartigraumaufde rsuchenachdiesternwelshegehabtbewohnbarplanetenkreisedrehensichund wohinderneurassevanverstandigmenshlichkeittkonntevortpflanzenundsiche rfreunanlebenslamdlichfreudeundruhemitnichteinfurchtvorangreifenvon andererintlligentgeschopfsvonhinzwischensternartigraum”

dv… is it a photo of you?

or possibly “Red Wacky League Antlez Broke the Stereo Neon Tide Bring Back Honesty Coalition Feedback Hand of Aces Keep Going Captain Let’s Pretend Lost State of Dance Paper Taxis Lunar Road Up! Down! Strange! All and I Neon Sheep Eve Hornby Faye Bradley AJ Wilde Michael Rice Dion Watts Matthew Appleyard John Ashurst Lauren Swales Zoe Angus Jaspreet Singh Emma Matthews Nicola Brown Leanne Pickering Victoria Davies Rachel Burnside Gil Parker Freya Watson Alisha Watts James Pearson Jacob Sotheran-Darley Beth Lowery Jasmine Hewitt Chloe Gibson Molly Farquhar Lewis Murphy Abbie Coulson Nick Davies Harvey Parker Kyran Williamson Michael Anderson Bethany Murray Sophie Hamilton Amy Wilkins Emma Simpson Liam Wales Jacob Bartram Alex Hooks Rebecca Miller Caitlin Miller Sean McCloskey Dominic Parker Abbey Sharpe Elena Larkin Rebecca Simpson Nick Dixon Abbie Farrelly Liam Grieves Casey Smith Liam Downing Ben Wignall Elizabeth Hann Danielle Walker Lauren Glen James Johnson Ben Ervine Kate Burton James Hudson Daniel Mayes Matthew Kitching Josh Bennett Evolution Dreams”

Tau.Neutrino said:

It might be ”Pablo Diego José Francisco de Paula Juan Nepomuceno María de los Remedios Cipriano de la Santísima Trinidad Ruiz y Picasso”

I hope so.

A brussel sprout.

Visitor here thinks possibly burmese, vietnamese, chinese.

If it’s a side on view it’s not Bill Lawry.

Julia Gillard

Lee Harvey Oswald…

mollwollfumble said:

Visitor here thinks possibly burmese, vietnamese, chinese.

He’s now said: “Pol Pot”

diddly-squat said:

dv… is it a photo of you?

It is someone even more famous than me

16 × 16

dv said:

16 × 16

it’s a scientist.

sarahs mum said:

dv said:

16 × 16

it’s a scientist.

Or an unwrapped mummy.

Bubblecar said:

sarahs mum said:

dv said:

16 × 16

it’s a scientist.

Or an unwrapped mummy.

no. a scientist.

sarahs mum said:

no. a scientist.

Which one?

It is clearly pluto…

• Which one?

Enrico Fermi…

Bubblecar said:

sarahs mum said:

no. a scientist.

Which one?

squint and look a bit side on.

sarahs mum said:

Bubblecar said:

sarahs mum said:

no. a scientist.

Which one?

squint and look a bit side on.

Out with it then…

dv said:

16 × 16

Not Tesla. Not Oppenheimer. Someone vintage.

sarahs mum said:

dv said:

16 × 16

Not Tesla. Not Oppenheimer. Someone vintage.

Spent too long in the bath. Archimedes.

It’s a chap with receding dark hair, but I can’t put a name to him.

dv said:

16 × 16

Hmmm.. blocky head, angular features..

Frankenstein’s monster..

.

.

.

.

.

or Robin Williams

Lee Harvey Oswald

Ian said:

dv said:

16 × 16

Hmmm.. blocky head, angular features..

Frankenstein’s monster..

.

.

.

.

.

or Robin Williams

More Tom Hanksish than Robin Williamsish.

Dr Smith.

sarahs mum said:

Ian said:

dv said:

16 × 16

Hmmm.. blocky head, angular features..

Frankenstein’s monster..

.

.

.

.

.

or Robin Williams

More Tom Hanksish than Robin Williamsish.

Is that a toothbrush mustache?

It’s probably some popular culture person I’ve never heard of.

Bubblecar said:

It’s probably some popular culture person I’ve never heard of.

Probably but that’s not specific enough for me to pay.

Is it front on or side on?

21 × 21

dv said:

21 × 21

Kevin Kostner

Peak Warming Man said:

Is it front on or side on?

That singing bloke who just died. Gay, weird.

• Dr Smith.

Maybe Mr Smith…

• Maybe Mr Smith…

Or Agent Smith, rather…

Peak Warming Man said:

dv said:

21 × 21

Kevin Kostner

Alan Turing.

Sweet Fanny Adams.

Maxwell Smart. Agent 86.

Prizes will only be paid on the mention of the person’s real name.

• Prizes will only be paid on the mention of the person’s real name.

Hugo Weaving then…

It’s an American actor but his name eludes me.

dv said:

Prizes will only be paid on the mention of the person’s real name.

Gordon Sumner.

furious said:

• Prizes will only be paid on the mention of the person’s real name.

Hugo Weaving then…

Don Adams

furious said:

• Prizes will only be paid on the mention of the person’s real name.

Hugo Weaving then…

You’re a fucking gun at this.

I was thinking it would probably either go this one or the next.

Ben Affleck.

Bubblecar said:

Ben Affleck.

I bet you don’t know what Ben Affleck looks like…

dv said:

Prizes will only be paid on the mention of the person’s real name.

Don Adams then?

dv said:

Bubblecar said:

Ben Affleck.

I bet you don’t know what Ben Affleck looks like…

I’ve just Google Imaged him and it could conceivably be him.

dv said:

Bubblecar said:

Ben Affleck.

I bet you don’t know what Ben Affleck looks like…

I’ve just Google Imaged him and it could conceivably be him.

• I was thinking it would probably either go this one or the next.

When you got to 16 × 16 I thought it would go after the next one because when you zoom out it becomes less pixelated and I figured the next one would be a lot clearer – though I was guessing your next one would have been 32 × 32 but that would have been a dead giveaway…

Well I was never going to get it since I don’t know Hugo Weaving from Adam.

dv said:

hugo weaving.

Bubblecar said:

Well I was never going to get it since I don’t know Hugo Weaving from Adam.

He’s for a jig or a tale of bawdry, or he sleeps.

Bogsnorkler said:

hugo weaving.

For those not quite following the ball, furious won.

Bogsnorkler said:

hugo weaving.

I think Boris should be awarded “Most Confident Answer”.

dv said:

Bogsnorkler said:

hugo weaving.

For those not quite following the ball, furious won.

i did actually say ‘or someone else’ early on in the piece.

dv said:

Bogsnorkler said:

hugo weaving.

For those not quite following the ball, furious won.

What does he get?

roughbarked said:

dv said:

Bogsnorkler said:

hugo weaving.

For those not quite following the ball, furious won.

What does he get?

The car.

dv said:

they should get him to play Oppenheimer.

Give us another one DV.

I never got to see it after 11 × 11.

Didn’t think I would ever pick this person as a favourite of the week, it’s Lorde, I thought it was Sara Blasko and worse than that it has a gay dance club vibe. But I have been playing it to death.

https://m.youtube.com/watch?v=dMK_npDG12Q

sarahs mum said:

dv said:

they should get him to play Oppenheimer.

or Lee Harvey Oswald

I agree with the above.
16 × 16 is not good enough for me to get any more than middle aged man with receding hairline and no facial hair – could even be Tom Hanks.
And with 21 × 21 it’s possible to tell Kevin Costner – just.

Let’s see what 16 × 16 looks like in full screen Irfanview. Hmm, nose still looks really strange. Why?

Because at 16 × 16 the nose is still completely missing – it’s masked by the darkness on either side. Or to pit it another way, at 16 × 16 the pixel boundary is right down the centre of the nose. At 21 × 21 the nose is present, because the centre of the nose coincides with the centre of a pixel.

One conclusion of the Scientific American study of pixellated faces is that if the pixels are allowed to be shifted a sub-pixel distance sideways and vertically during the image generation process, so that specific features such as eyes and in this case nose fell in the middle of a pixel, then identification becomes much easier even at lower pixel resolutions.

What software did you use to produce the pixellated images? Can you produce a 16 × 16 image with the pixels shifted a half pixel width sideways?

Is it over?

mollwollfumble said:

And with 21 × 21 it’s possible to tell Kevin Costner – just.

It is not Kevin Costner.

What software did you use to produce the pixellated images?

Gimp pixelize, which takes a simple arithmetic mean of values within the square.

kii said:

Is it over?

It is over

dv said:

kii said:

Is it over?

It is over

No way is it over. I have a new collection of about 20 more celebrities to try on you.
Try this. 3*3

mollwollfumble said:

No way is it over. I have a new collection of about 20 more celebrities to try on you.
Try this. 3*3

Dick Van Dyke.

Bubblecar said:

mollwollfumble said:

No way is it over. I have a new collection of about 20 more celebrities to try on you.
Try this. 3*3

Dick Van Dyke.

That has a striking resemblance to Roger Braintree.

mollwollfumble said:

dv said:

kii said:

Is it over?

It is over

No way is it over. I have a new collection of about 20 more celebrities to try on you.
Try this. 3*3

It’s Charles Bronson. I took a while to pick it because he is smiling in the shot, but I eventually worked it out.

Peak Warming Man said:

Bubblecar said:

mollwollfumble said:

No way is it over. I have a new collection of about 20 more celebrities to try on you.
Try this. 3*3

Dick Van Dyke.

That has a striking resemblance to Roger Braintree.

I’d say he’s flicking the bird.

I’m limiting it to four celebrities. 5 × 5.

Celebrity A.

Celebrity B

Celebrity C

Celebrity D (no 5 × 5 image)

mollwollfumble said:

I’m limiting it to four celebrities. 5 × 5.

Celebrity A.

Celebrity B

Celebrity C

Celebrity D (no 5 × 5 image)

A) Dick Van Dyke
B) Bill Cosby
C) Penelope Keith
D) Dolly Parton

Bubblecar said:

A) Dick Van Dyke

Affectionately know as Penis Van Lesbian.

Woodie said:

Bubblecar said:

A) Dick Van Dyke

Affectionately know as Penis Van Lesbian.

Remarkably Mr Van Lesbian is still alive, aged 91.

Woodie said:

Bubblecar said:

A) Dick Van Dyke

Affectionately know as Penis Van Lesbian.

:)

One Nation

Bubblecar said:

mollwollfumble said:

A) Dick Van Dyke
B) Bill Cosby
C) Penelope Keith
D) Dolly Parton

Not yet. Please come again.

OK, 5 × 5 again on Celebrity A, new 7 × 7 on B, C and D

A. B.

C. D.

mollwollfumble said:

OK, 5 × 5 again on Celebrity A, new 7 × 7 on B, C and D

A. B.

C. D.

B = Tom Baker.

sibeen said:

B = Tom Baker.

Not there yet.

Four new images, 7 × 7 on Celebrity A, 9 × 9 on B, C, and 13 × 13 on D

A. B.

C. D.

You might have a 50-50 chance of identifying them at this resolution.

mollwollfumble said:

Four new images, 7 × 7 on Celebrity A, 9 × 9 on B, C, and 13 × 13 on D

A. B.

C. D.

You might have a 50-50 chance of identifying them at this resolution.

Mickey Mouse, Sharbat Gula, Che Guevara, Mardy Feldman

mollwollfumble said:

Four new images, 7 × 7 on Celebrity A, 9 × 9 on B, C, and 13 × 13 on D

A. B.

C. D.

You might have a 50-50 chance of identifying them at this resolution.

They all look like robots.

D = Groucho Marx.

As was the case with Mona Lisa, I mainly worked out Gula by the picture’s colours.

dv said:

As was the case with Mona Lisa, I mainly worked out Gula by the picture’s colours.

NHO Gula. You’re probably right with the others.

dv said:

mollwollfumble said:

Four new images, 7 × 7 on Celebrity A, 9 × 9 on B, C, and 13 × 13 on D

A. B.

C. D.

You might have a 50-50 chance of identifying them at this resolution.

Mickey Mouse, Sharbat Gula, Che Guevara, Mardy Feldman

Bingo. 100% correct. Well done!

A. B.

C. D.

So, in summary, all are faces:
Hollywood star (B/W) in 21 × 21
Famous painting in 7.2 × 10.3 (or better)
Cartoon in 7 × 7
Famous photograph in 9 × 9
Iconic stencil in 9 × 9
Comedian in 13 × 13

I wonder if I can do better with that famous painting, I just took that one straight off the web. Let’s see.

mollwollfumble said:

…
Famous painting in 7.2 × 10.3 (or better)

I wonder if I can do better with that famous painting, I just took that pixellation straight off the web. Let’s see.

6 × 9 (homemade), 7.2 × 10.3 (straight off web), full scale

I think it’s as easy to recognise from the 6 × 9 as it is from the 7.2 × 10.3 ?

mollwollfumble said:

dv said:

mollwollfumble said:

Four new images, 7 × 7 on Celebrity A, 9 × 9 on B, C, and 13 × 13 on D

A. B.

C. D.

You might have a 50-50 chance of identifying them at this resolution.

Mickey Mouse, Sharbat Gula, Che Guevara, Mardy Feldman

Bingo. 100% correct. Well done!

A. B.

C. D.

So, in summary, all are faces:
Hollywood star (B/W) in 21 × 21
Famous painting in 6 × 9
Cartoon in 7 × 7
Famous photograph in 9 × 9
Iconic stencil in 9 × 9
Comedian in 13 × 13

So, Mickey and Mona are equally recognisable. I didn’t expect that, I thought Mickey would be easier.

Follow up question. What’s the minimum number of bits of information needed for facial recognition?

I can’t just use the file size, because dv’s 16 × 16 jpeg of Weaving is only half the size of my 3 × 3 jpeg of Mickey.

mollwollfumble said:

I can’t just use the file size, because dv’s 16 × 16 jpeg of Weaving is only half the size of my 3 × 3 jpeg of Mickey.

Surely it’s just number of squares x number of shades or colours.

The Rev Dodgson said:

mollwollfumble said:

What’s the minimum number of bits of information needed for facial recognition?

I can’t just use the file size, because dv’s 16 × 16 jpeg of Weaving is only half the size of my 3 × 3 jpeg of Mickey.

Surely it’s just number of squares x number of shades or colours.

Compressing to jpeg 1%, the smallest Irfanview will allow, gives heavily pixellated but easily recognisable faces. Che still looks remarkably good.

Weaving 880 Bytes and Mona 570 Bytes

Mickey 670 Bytes and Afghan girl 530 Bytes

Che 740 bytes and Marty 890 Bytes

> Surely it’s just number of squares x number of shades or colours.

Let’s see if that formula matches what I get above. Weaving, 8 shades of grey (and quadratic graded colour but let’s skip that) 32 × 32 pixels. 8*32*32 = 8192 bits, at 8 bits per byte that’s 1024 bytes, which is definitely in the same ballpark as 880 bytes, multiple squares with the same shade would help file size.

mollwollfumble said:

What’s the minimum number of bits of information needed for facial recognition?
…
530 Bytes, 880 bytes.

I wonder if we can do better than that for both file size and recognisability by applying in turn greyscale (three images) resize, jpeg 1% then view fullscreen and screendump.

The first identikit, developed in 1959, had 500 mix and match components. If I make the assumption that those components were separated among 10 features (face shape, hair, hair length, ears, mouth, nose, etc.) then that’s about 10^(500/10) = a mere 150 bits of information. Big gap between 150 bits and 530*8 = 4240 bits.

mollwollfumble said:

mollwollfumble said:

What’s the minimum number of bits of information needed for facial recognition?
…
530 Bytes, 880 bytes.

I wonder if we can do better than that for both file size and recognisability by applying in turn greyscale (three images) resize, jpeg 1% then view fullscreen and screendump.

The first identikit, developed in 1959, had 500 mix and match components. If I make the assumption that those components were separated among 10 features (face shape, hair, hair length, ears, mouth, nose, etc.) then that’s about 10^(500/10) = a mere 150 bits of information. Big gap between 150 bits and 530*8 = 4240 bits.

But if a “bit” is a complex shape that occurs in faces it actually contains much more than one bit of information.

A related question is how many notes do you need to identify a piece of music. It depends on the piece of course, but there are certainly pieces that can be identified after just 3 notes, but only because each note contains much more information than just pitch.

The Rev Dodgson said:

But if a “bit” is a complex shape (identikit) that occurs in faces it actually contains much more than one bit of information.

A related question is how many notes do you need to identify a piece of music. It depends on the piece of course, but there are certainly pieces that can be identified after just 3 notes, but only because each note contains much more information than just pitch.

OK, on identikit. The brain is a non-linear device. After applying squinting, it doesn’t really care much whether the “bit” is physically encoded as a pixel or a wavelet.

My idea of viewing full screen and screendumping fell through, it just highlights discontinuities. It fails miserably for heavily compressed jpegs. Another idea that failed was reducing the number of colours in the palette before compressing – has a negligible effect. Switching to grayscale during jpeg compression saved about 25% of space on the greyscale images.

On music, yes. I can think of one piece of music identifiable by just the first “note”, although in this case the “note” is a discord with several pitch components as well as timbre components.

Please. Recognisable at what level of resolution?

From top left, 210 bytes, 250 bytes, 330 bytes, 33190 bytes.

Please. Recognisable at what level of resolution?

From top left, 210 bytes, 250 bytes, 330 bytes, 33190 bytes.

192 kb image for me.

mollwollfumble said:

Please. Recognisable at what level of resolution?

From top left, 210 bytes, 250 bytes, 330 bytes, 33190 bytes.

The 330 byte is surprisingly recognisable when squinted at.

You could generate a recognisable picture of HW for under a kilobyte, using scalar vector graphics.

This file is 615 bytes, just using JPEG technology.

dv said:

This file is 615 bytes, just using JPEG technology.

Oh, that’s whatisface!!!

dv said:

This file is 615 bytes, just using JPEG technology.

For me, that’s recognisable as a person, likely male (receding hairline), but not a specific person.

Michael V said:

dv said:

This file is 615 bytes, just using JPEG technology.

For me, that’s recognisable as a person, likely male (receding hairline), but not a specific person.

Kevin Costner!

Or that guy who was married to my friend Cath.

mollwollfumble said:

Please. Recognisable at what level of resolution?

From top left, 210 bytes, 250 bytes, 330 bytes, 32190 bytes.

Better or worse? Are the images at right easier or harder to recognise than the images at left?
The images at right have to same number of bits as the images at left, but they have bigger pixellation with more subpixel resolution.

mollwollfumble said:

mollwollfumble said:

Please. Recognisable at what level of resolution?

From top left, 210 bytes, 250 bytes, 330 bytes, 32190 bytes.

Better or worse? Are the images at right easier or harder to recognise than the images at left?
The images at right have to same number of bits as the images at left, but they have bigger pixellation with more subpixel resolution.

Top row equally unrecognisable.
Bottom right more recognisable than bottom left for me.

mollwollfumble said:

mollwollfumble said:

Please. Recognisable at what level of resolution?

From top left, 210 bytes, 250 bytes, 330 bytes, 32190 bytes.

img src=“http://i120.photobucket.com/albums/o162/DavidPaterson/SSSF/weaving4_zpsya7slzr4.jpg” height=“600” width=“600”

Better or worse? Are the images at right easier or harder to recognise than the images at left?
The images at right have to same number of bits as the images at left, but they have bigger pixellation with more subpixel resolution.

img src=“http://i120.photobucket.com/albums/o162/DavidPaterson/SSSF/weaving4a_zpsl0wp58di.jpg” height=“600” width=“600”

Right is better, but not recogniseable by me as a specific person.

Does milk go stale or off?

Sorry.

Witty Rejoinder said:

Does milk go stale or off?

Yes

Witty Rejoinder said:

Does milk go stale or off?

Yes. Commonly bacteria.

> You could generate a recognisable picture of HW for under a kilobyte, using scalar vector graphics.

Scalar vector graphics? You mean like a cartoon?

dv said:

This file is 615 bytes, just using JPEG technology.

How did you do it?

You can see jpegs by me at 320 bytes above and 880 630 bytes at fifteen or so posts further back. The reduction from 880 to 630 was switching to greyscale.
But mine are heavily pixellated.

For the 320 byte one I cut image size by four to 64 × 64 then saved at 8% jpeg greyscale.
For the 630 byte one I saved full size image 256 × 256 as 1% jpeg greyscale.

The 320 byte doesn’t seem recognisable as Weaving but the 630 byte one does.

>> You could generate a recognisable picture of HW for under a kilobyte, using scalar vector graphics.

> Scalar vector graphics? You mean like a cartoon?

I don’t understand why the .svg files of Che Guevara on the web are so large. Two .svg files I downloaded have sizes 29 and 125 kB.

mollwollfumble said:

>> You could generate a recognisable picture of HW for under a kilobyte, using scalar vector graphics.

> Scalar vector graphics? You mean like a cartoon?

I don’t understand why the .svg files of Che Guevara on the web are so large. Two .svg files I downloaded have sizes 29 and 125 kB.

Some people aren’t very good at it.

Using a somewhat different methodology, this one is 597 bytes.

dv said:

Using a somewhat different methodology, this one is 597 bytes.

580 bytes (or so) as a png, 1150 bytes as a gif.

I’m surprised, I normally save BW images as giff.

dv said:

Using a somewhat different methodology, this one is 597 bytes.

That’s very close to specifically identifiable.

It looks as if 500 bytes +- 100 bytes really is the limit for face recognition.

It is not really a fair test any more since we are all primed to see Hugo.

But here is someone else, coming in at 965 bytes. I think about a third of that is for the hair, but I must admit that helps with the identification.

Can y’all identify?

Elon Musk.

dv said:

It is not really a fair test any more since we are all primed to see Hugo.

But here is someone else, coming in at 965 bytes. I think about a third of that is for the hair, but I must admit that helps with the identification.

Can y’all identify?

It’s PWM.

Oh we got some comedians in the house tonight.

dv said:

It is not really a fair test any more since we are all primed to see Hugo.

But here is someone else, coming in at 965 bytes. I think about a third of that is for the hair, but I must admit that helps with the identification.

Can y’all identify?

Could be Steve Bannon after a diet and haircut.

Bubblecar said:

dv said:

It is not really a fair test any more since we are all primed to see Hugo.

But here is someone else, coming in at 965 bytes. I think about a third of that is for the hair, but I must admit that helps with the identification.

Can y’all identify?

Could be Steve Bannon after a diet and haircut.

His skin looks kind of pixelated in real life

kii said:

dv said:

It is not really a fair test any more since we are all primed to see Hugo.

But here is someone else, coming in at 965 bytes. I think about a third of that is for the hair, but I must admit that helps with the identification.

Can y’all identify?

It’s PWM.

robert downey jr?

stumpy_seahorse said:

robert downey jr?

Quite so

Bubblecar said:

dv said:

It is not really a fair test any more since we are all primed to see Hugo.

But here is someone else, coming in at 965 bytes. I think about a third of that is for the hair, but I must admit that helps with the identification.

Can y’all identify?

Could be Steve Bannon after a diet and haircut.

Could be…or it might be no one at all.

dv said:

stumpy_seahorse said:

robert downey jr?

Quite so

Didn’t get that.

But then I wouldn’t have recognised him with near infinite bits either.

The Rev Dodgson said:

dv said:

stumpy_seahorse said:

robert downey jr?

Quite so

Didn’t get that.

But then I wouldn’t have recognised him with near infinite bits either.

The Rev Dodgson said:

dv said:

stumpy_seahorse said:

robert downey jr?

Quite so

Didn’t get that.

But then I wouldn’t have recognised him with near infinite bits either.

I mean no offence but what the fuck is wrong with you people…

dv said:

The Rev Dodgson said:

dv said:

Quite so

Didn’t get that.

But then I wouldn’t have recognised him with near infinite bits either.

I mean no offence but what the fuck is wrong with you people…

dv said:

The Rev Dodgson said:

dv said:

Quite so

Didn’t get that.

But then I wouldn’t have recognised him with near infinite bits either.

I mean no offence but what the fuck is wrong with you people…

Both gay and lame as we used to say in the old place.

dv said:

The Rev Dodgson said:

dv said:

Quite so

Didn’t get that.

But then I wouldn’t have recognised him with near infinite bits either.

I mean no offence but what the fuck is wrong with you people…

I’m with the Rev, and the reason would be that I’ve probably never encountered this actor in anything I’ve seen, and I don’t read celebrity pages.

dv said:

The Rev Dodgson said:

dv said:

Quite so

Didn’t get that.

But then I wouldn’t have recognised him with near infinite bits either.

I mean no offence but what the fuck is wrong with you people…

Their hopless.

Looking through this list, I have never seen any of these films, and doubt I ever will as they don’t seem to appeal.

https://en.wikipedia.org/wiki/Robert_Downey_Jr._filmography

Okay, how about this one at 989 bytes.

dv said:

dv said:

Okay, how about this one at 989 bytes.

dv said:

Easy, Boy George.

Bubblecar said:

dv said:

The Rev Dodgson said:

Didn’t get that.

But then I wouldn’t have recognised him with near infinite bits either.

I mean no offence but what the fuck is wrong with you people…

I’m with the Rev, and the reason would be that I’ve probably never encountered this actor in anything I’ve seen, and I don’t read celebrity pages.

You didn’t even see Chaplin?

Peak Warming Man said:

dv said:

The Rev Dodgson said:

Didn’t get that.

But then I wouldn’t have recognised him with near infinite bits either.

I mean no offence but what the fuck is wrong with you people…

Their hopless.

Eye neau

dv said:

Okay, how about this one at 989 bytes.

dv said:

Judy Garland.

Peak Warming Man said:

dv said:

Okay, how about this one at 989 bytes.

dv said:

Easy, Boy George.

I don’t think we are in kansas anymore..

dv said:

Bubblecar said:

dv said:

I mean no offence but what the fuck is wrong with you people…

I’m with the Rev, and the reason would be that I’ve probably never encountered this actor in anything I’ve seen, and I don’t read celebrity pages.

You didn’t even see Chaplin?

No.

dv said:

Okay, how about this one at 989 bytes.

dv said:

Looks like Anne Briggs, but I doubt that it is.

dv said:

The Rev Dodgson said:

dv said:

Quite so

Didn’t get that.

But then I wouldn’t have recognised him with near infinite bits either.

I mean no offence but what the fuck is wrong with you people…

I’m saying nothing.

768 bytes.

dv said:

Okay, how about this one at 989 bytes.

dv said:

Princess Leia, when she lets her hair down.

dv said:

768 bytes.

Alfred Hitchcock.

dv said:

768 bytes.

If you turn it upside-down it looks like that clown dude.

Bubblecar said:

dv said:

Okay, how about this one at 989 bytes.

dv said:

Judy Garland.

Correct

Michael V said:

dv said:

768 bytes.

Alfred Hitchcock.

Correct

I give up.

Hitch was too easy :)

Bubblecar said:

Hitch was too easy :)

Yeah, even I got that one.

The Rev Dodgson said:

Bubblecar said:

Hitch was too easy :)

Yeah, even I got that one.

Heard of him have you?

> 965, 989, 765

Files of identifiable faces are no smaller in this sort of cartoon format than in jpeg.

I had previous identifiable faces in raster (admittedly only just identifiable) with
570 bytes – Mona Lisa (also 560 bytes, not posted)
630 bytes – Hugo Weaving
670 bytes – Mickey Mouse
530 bytes – The Afghan girl
560 bytes – Che Guevara
600 bytes – Marty Feldman

Can you break the 500 byte limit with a real face?

dv said:

The Rev Dodgson said:

Bubblecar said:

Hitch was too easy :)

Yeah, even I got that one.

Heard of him have you?

I’ve probably even seen one or two of his works.

On the Bairdboard bombardment screen.

dv said:

The Rev Dodgson said:

Bubblecar said:

Hitch was too easy :)

Yeah, even I got that one.

Heard of him have you?

I’ve probably even seen one or two of his works.

On the Bairdboard bombardment screen.

The Rev Dodgson said:

Bubblecar said:

Hitch was too easy :)

Yeah, even I got that one.

I’d have said it was some dead person. Looks dead…

Okay well since those last two were easy…
966 bytes.

You need to do someone I might know…

F’r‘instance…

dv said:

Okay well since those last two were easy…
966 bytes.

He wrote The Moon is a Balloon.

mollwollfumble said:

> 965, 989, 765

Files of identifiable faces are no smaller in this sort of cartoon format than in jpeg.

I had previous identifiable faces in raster (admittedly only just identifiable) with
570 bytes – Mona Lisa (also 560 bytes, not posted)
630 bytes – Hugo Weaving
670 bytes – Mickey Mouse
530 bytes – The Afghan girl
560 bytes – Che Guevara
600 bytes – Marty Feldman

Can you break the 500 byte limit with a real face?

All of these images that I’ve presented are real faces. I’ve used contrast and edge detection and I’ve blanked the background but they are all just filtered photographs.

But you’ve set me a challenge, I’ll try to go under 500 bytes for a photograph that is not famous. I’ll avoid obvious ones like Monroe, Groucho etc.

Peak Warming Man said:

dv said:

Okay well since those last two were easy…
966 bytes.

He wrote The Moon is a Balloon.

David Niven.

Peak Warming Man said:

dv said:

Okay well since those last two were easy…
966 bytes.

He wrote The Moon is a Balloon.

You’re good, you’re very good, PWM.

>>I’ll try to go under 500 bytes for a photograph that is not famous

Yeah your neighbour or someone at the bus stop.

dv said:

Peak Warming Man said:

dv said:

Okay well since those last two were easy…
966 bytes.

He wrote The Moon is a Balloon.

You’re good, you’re very good, PWM.

His era.

What about this one
data:image/jpeg;base64,/9j/4AAQSkZJRgABAQAAAQABAAD/2wCEAAkGBxMTEhUTExMWFhUXGR0YGBcYGRsdGhodGh8YHR8dGhgaHSggHholGx8dITEiJSktLi4uGh8zODMtNygtLisBCgoKDg0OGxAQGi0lHSUrLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLf/AABEIALEBHAMBIgACEQEDEQH/xAAcAAACAgMBAQAAAAAAAAAAAAAFBgMEAQIHAAj/xABIEAABAgQEAwUFBQUFBwQDAAABAhEAAwQhBRIxQQZRYRMicYGRMqGxwfAHI0JS0RQzkuHxFWJyssIWNHN0goOzJENFYxclNf/EABoBAAIDAQEAAAAAAAAAAAAAAAECAAMEBQb/xAAoEQACAgICAgICAQUBAAAAAAAAAQIRAxIhMQRBIlETMnEUM1KBkUL/2gAMAwEAAhEDEQA/AHT/AGkQpTB4sHF0jV/fHPKXEOzXmdw94f8ACK+TUpDM8InY7jRucZl/Txg4tL5++CSsNRyHpGgwxH5R6QQFGmqQpbpeC+Y8zEcikSnQRYIggAfE6VKlEC8c/wAWpV5EpEony09I6PjqiJZIhQlYirdIPhCuai+Rowb6E39nWnZaTyuIgqD3SY6AitC+6UkP5wn8SYcZTqCnCi+jNDRkn0FprsG8Nq++T4×22nJyJudI4dw0fvUn+9HbqZf3afCLYlcicE8z6×4vGoMbJIdL6KLDrYn4A+kM3SsVcujREtxzjRdEDqItLUUpKmJAsEgEnoGiH+0mH3iGDsCC5ALsSGBAe3R4olninTLo4mwfNwRCtm6xHNwAN3VEesXqrF5ctSXu9rKDjkSCXIJ5OekWZNciYgLQQR8PlruNdnhfywYfxSF2Zhk5PsrMQibUI2JhtCg2x+AjQSHDt8×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×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×5PC3W8VTc/384Spbt91LUpjyfnbSB8p8EdQLXEOFTEIZXdv7R9lxo+UWHUaQuSESykJXLQyEqSkIWVDOopdalfiITnLX2hmkYoiakolz1zEkaTE5bHcAjT339VLHVdksJzZBqFgG5GxF7EPpbpqYfFd6iZarYmwKoAQpKpQe3fUC4B/KOXhHRuGMQShEw08tSwE3WWKlqSbDYMxPWEGWrtZQUt0kJ7qQRlmN+IpKn3bu22jGEVyynJLQFIF1m9sxZidByA98NJNttCKqSOhcRk19HmSDLmBz2ZLv4EgPpq2×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×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×8JSKelmMSe82oGu0UcKkTJkxAEtYlSQQJl8pId1E7kn4RBT0qVDMUAHMpzzH9IRwhrFVyM8k1OTvgF4bQsxWNLub36xcn4jdgWAjTEpgSDp198BVTuusN2VhairlpW6VEh/q/1pDdhFAMSmBUyaQUBgm3n9COfU9S2/wBeMHMExYyZ6FJ3LEc30+UH2Q6vR8NyZIZ3bnFkyJOhSIohVQoPlAB0ciIjh1QovnSnwc/pE0j3Q28vsuzpckfgDxhNZKAbKn3RWOBqPtzleQb4mN5WAIG61ef6CCoL0gPJJ9yZujEUA6CN51cnUM0T0vD8l3MsqHUq/WJzgKNpQA6/zMMDlg9GLpZnEYGJJ5iCX9hAfgR6CJ5WCjkjXkP0iKxWaGUHPjGpSIkWLmNMpirUssrYhUTJcpSpSilXT9DaOQD7TMX/AGgyO2BOfKxlo56+yNrx2ObIKg2xgBM4KklfaMM3OClQLI0rxRaAoz0×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×8+TdiT0GEYHh4LhJJN7qVfyBiCsrKdLgAJSnlCNmmCyiv66vGUys+gUX3OnpFZbRZrK/MpWTSLVHTtJWOaFX8QREdJhtw+0XsTmdlTzV8kkAczt7zAbvoaKo5PxHg5pakyge7lRMQS10zEpUPFnbyh04f4TRUpQuVOKXCc0lCAZg1dipYGTXvAEszgbw/bFR9lXSU7ikkv4pK0W/hiphxCpYTvtdiDqG87ahrefTcm4pnIS5Y38TcR11HTBMqmRKpUNLaaAqasXZSk91gWNm8es1MJVZTEyky01CUFQQgsVsHKcj3UbsUxzPHBPKsswkgPucvKwJtfaB9IFO2bzPzg62rBfoNYlhtat1ikqcg1PYTWtq5ywG7WwgthWJLlTE5VqQXspCin1ys490NieG5GJFUxNSJNSLzR2YIWAPbyAp77aqSSCzsCSSVKuwUIlPM+v5xiZOOdBB0UnTWx0hoxXhekpkgrxNJUfwinU/i3aO2m3rC7OwpSJ9OlwtM1aMkxDlCwVAEpJALjQpIChuBBTTAfUMikQEIcP3Q3pu0TIko/KInCDlA5ARqpMM2REa0bAD0jTJd4laMNAuh9UzIjwMYePQBqoyrSKqFa+MWTFiVTpbSHi0iuasGmckbRtLmJO0Zn0aEi5v1inLAGhisIUGQBy0UJpBLiNTeMhAgkMARNLIuSwAuTt4xGvIlJUoskXJMIvG/FTyFy5QKAvuk/iIGvgP0MK3RErB+N8QmurBLlH7qW9ue5zHq20DsVqgh3U7agabtz09bdYBS640tEtaCRMnKKQTZkJuW8T4WG8VFrVM7FIVdaQSTs9/hf+l6qvkcMyp5U5dkjmT5WJLaMf6MHqq1c1WVDqJ7oAFy+gAGp5b6xbrR93lStMuSmypq3GYixSgBLzFa91O+pF4FzatMtBSgKlJI7yyQJ80HUWcSUEWKQSSzFRgxiRs9+6UQFhdQCQ4Lyqc2YuLTJ45B0pPMjuluGgJQmzhcISSFG5MyY6QTzPtF7XSNYVqZSpqkpQkBLsGGnLb+ekNnFCP2WmTTgjOSFzAG1YhKeeno/N4aX0CL9idiExU6ZkT3lqUEpY3KlFm8zHXPtMRlSikl2RLSinQBoBlGY/wAAbzjm3A1J2uKUSSbduhXjkOc+uVo6txhKK64Ol2zKI5ZhLAbr3Yq8l0kW+Ktpsr8DUQpy7ahj7oOYrTF8w0v9eMa0VN3bbxalzj7KtowXfZ0aFKppS728/wCUayKTkG6c/rnB+tp3MRypECwlSVT5Rfx/rFBUlVTVU9MkWMxK19ESyFKfoWy+KhFzFagaDeLnDtRKoaOoxKddV0ID65PZQOqplvIcosxR2kJmnrA5v9rGIdvi07lKCZIb+6HPnmUr0ijSqIYMWF7N15/GAParmLVMWXUtRWpR3KyST6wxUiwUsRoAzf0/nHSkqSRyY8sixz7xAULHla+2w+jC9QqyzQBaGirlgo0Z/c+31/VSz/e9QWPrrDQ5TQJdhGtcEF2izJq1JAWCyhoQoght9XiCsSFAH6+EU1XR7oKVoHRbxbiKfUpCZh7RvZUQkqYf3mze+DnBfFqqdBlhFtznUFAjcEukfw/rASkQlKbi/wDS3OKkghMw/R+ngtJqkQ6nJ+1SelViSn8q2Ud92c6bHeG7/wDKlNLpBOnJOcnLll3B3Bd7DXnoY4Ash9ngvhNXLCSienOhVlDfxBcOoWN/hC1QUzreHfaiZqTM7FARsHLnz/lBCR9qVE7TcyOoIUPPQxyKjUJMjKiyVEqN7sdBccm6XgGjCv2iYQCEHd9B5wF32NZ9QYZjdLU3kT5cw/lB738JvF9Qj5xwrApVOoEiYuY4Y9pkQ+lslyfEt4QwYt9ocyVlkS5iyRu5J/iU6m8SYOxDs65sWJU8tpHDZXHVQlQeaog8yCPfpDjhvGEwodUtKjzBKfdDbRFaY1zS5Lm7xoBES3c+Mby0EloUhKmI6+slyEGZOWEpHPU+A3hdx3jKVImdhKaZN3a4T0tqfq8cz+0PG5sxQkknPMIB21Zh4QNvSDQ18QcWKqVU6EpKJS1BYFyVJ2JIs+jDqfGFLFKztVLOgTYB9g/xPxiTFKnLiCUo0lshOgskBIB6Ns+haFisqFCaq+p9/nvCJNsZtIs8Q1QCJaEn2B/N9YtUeKS5FPLmMFz1JCUBYdCAnK6yPxFywTp3S7vCzVrJMShbyQNdr8gXt4n4RbpwhNuQjMxEqV2q5qZkzZS1Pl6JSWSEjkBEEuTnXmWoLP8AiT81QOpEPmDaJJ9IpqhlAFj1Q1lPSI7ZZzTNUy0LS7tZyl2AOpN+QgTPxJVSVzlICEp5Em/K/iNekAainysCO9v57QbwKnVMkmXLTmmKmME7lwlgOd/fCuCSsO18Bz7KZWfF6csfu8yj/CR846xxAWxFT3CkI/1RzX7GJCjiaSNkEl/y3BYc7NHQuKS2IqbRMpHh+LTpeMnk9Gnxf2/0GqcXDb/N/mPhGainCrixiKlBygvb5X+UWJa7NvGWjeUKqSXHuiBSmQSByghVTOY0+cVVqCv8P62HzhGhhfkYYuompQkZQTdR/CNSfIfGAX2v12WWmmSMssKyy0DYJuVK6nMSddUvd46JUThLlqWklJUyE5SkHXve0CwskP1jjXFqDPnqU47NCciFH+Jcw+Kn8gOcavHhXJi8mbbaQtSVBEtzfT3P+oixTYshx3iPEAj0tFbHq6WpRRJSUSxoCSS3UnUneBBEdBQtWzDdDzJmicCAUnTTQOLEgsQIVsSpyib13Hhy6Mx84pyJ6kEFJYj69IKV1Z26AojvpuTzHTf9GiKOrI3aJKeozJy7i7RVIuobO8R0q+cWJywL7nf6+ESqZDVKydDGgXqTGiVsI1mqsxiUAzTqJufOJzMJsDbQxUTMYWjaWvKMx1OkFogTXWlDaWs2×8YtJxGVMA1QvS2hvuwcX5bQIlLJTcb2+ESyJYCtO9zhGkFMMzcTVJlFRU6iMqdxto+0AKKYpSipXeUrTdyY1xWpzEJBsIwibkS7XZhDKNIjdhRGMCnfIBMnn8ZYpR0SPzdTpA6Zi1SolXarUTqQTGuGYeqarkn8Sjp6wa/s2ULBc635SgD69YjpEVs+glVF/OFr7ReL/wBikZJf72YNd0p09TfyEFauaUJKmc6AdevSOM/aTULmT3VfY+Voru3Q3qyxwGO0mrqJqu7LGYlR1PJ+cAMVrzMrUrI0mJIBf8wOirtFvAMTEtPZAkOXtbbUHnoP4ucUVyXqktfvD49IkVUmRvhBPiSe1Uo313O/19XgXjCnVmG/8vhpE+PqeaTa3LT1MUpi7XgxXTBJ8sqzZjiJpf7oefxMU5mpibtPuwOpi1rgQnog0uYvmyR8T8omwmmypNQsBgSEA7qDOW5B/XwibD8NVMTLl6AgzFnkktc9coS3+IdYtYpNBOVIZCbJHIXs/q/WEcvQwKIclRuS8WaR0AEWKy/pobecYpqYzFhI3IHhFnE1BM1KdkkBvDWA36IvsevsUR/+wUr/AOhZPiVym9xPrD1xLLauUXbMhJfo2VvcfWEn7E5g/aqiYohKUyWLmw76fIWEdF4klhVQhQYgyhccsyiC+4jHn6NnjP5GlM+QjTnGhqWI31+Rfw1iSVItsCNDy+vGI58nUkufB+u8ZeTbwD6nEhnI3JHUDT6bp1j1JMzqCQ6nISBu+npFHFKfvv7mbwfZnfSLGGSuzkVNSvuhEpQSpnDhJzEi2zJf+8ryiVsk5qMbOf8A2kcaKXN7GnP3Mp0PutX4iCD7JU7NsPCEKpxKYtOVSi3ifmYwsFRcuSw994j7KOvCMYo40pNuyFIvE0yW0bCXeJpqIdyFKChElPNKTb+vSMzURolN4btELiAAbaGJJpGhjSXLsN4xUGE9hMGw+UV56r/KMqWbxokHWGSAbITz56RrNW56RkrtaIxBSIEKRi1yGvbnHpk5g++kVJa2jRayYXXklmHc3i5TUy5ymSCw6ge8xVkSStQSPMnQDmTyhgTPly0ZEzkAbkOST6WESToKRn+zlMAqYhIG2ZgPSKk6kD/7zLHmr9IKU+ECalRzLATd1S1JTt+IljtpA6dgZ/Cyhz158zFV/Y1HScN4yTUqH3uV/aJ0HiH/AEEIvElVLmz5i83dzHLd7OWJ5+Qirj2AqkEapP8AXSAObYwIY03aYZS9E00gNlU+4MHMNyFXbqIGUEgc1ch4a+ULqeXOJJCz7PnFso2hEy9Om5i53it2iQNdOUV5q9oigqILN1KeJqWTnUhF2Ju3Lf3CKzw3cHUWRJqVi57sryIdflt1BiTeqDFWwjOldlLKSwUouvyslAPJItydzC7NU7tz1/rBHEp4Pde2vn8AYq0UoLmJ5Av9GKI/Y8ueEE8KpuylmafaNkguLHrAeaCtdxr7oM49VkgIAsAwDbeAiLAsLmVM6VIkN2sx2KjZCbPMVvbYa+oYJ+wtejov2HYak/tsxizypdi10hSjcbuRDnxAsGqSC9pYv4qVF3g/hqTQSBIkub5lrOq1lnUfTTaF/E6kqrJlnTmyA8soAPk4jNnl8TT4y+YZSgRRrFjceHMHYA7P+sTSpmt9APB7n5QFqsSZTHZjfof1jOzYgPjZWqYESg8xZCEjYlR1caDe+jbbsPFlOEUCpCGbJkzPdTAnMR1Lq8WihwpSFa5lWoF0FUtA0uQM6h4DujxVyhll0naq7ydjqxZrML829DFkVSM2SVyr0fMclL7cj9GJuygjJwaYapVJLTmmCYZKQSwspSXKjZmDnpFniTh2dRTBLnGWSoZgqWvMktqHYEEOLNuI6Gxg1YDCLxrNDmLSNYrTdYKIRLTaKxSzGLihaIJo2+rQ6YCVEznGsxUYkaRhRvEIVt4yJjBo1XGph6AeJjwMZQgnQRewyr7FRV2UqaWIaanOkPuEuO91gsBQeMQfp8KVOkT6splS5UogFswJUrRMtL6+MVMCwlVRMOXupGqjoHOz6qZyB0hdkNqXMEpZCBmnpmTDr2UsaBn76trXyuOsHU8RSUMlNN2GxKEAbA3KQXt5wYRJlISESwG0A/M97l7uzknZ/wA1h1ZTpO9i5zG3cBda/BR06×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

kryten said:

What about this one
data:image/jpeg;base64,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×22nJyJudI4dw0fvUn+9HbqZf3afCLYlcicE8z6×4vGoMbJIdL6KLDrYn4A+kM3SsVcujREtxzjRdEDqItLUUpKmJAsEgEnoGiH+0mH3iGDsCC5ALsSGBAe3R4olninTLo4mwfNwRCtm6xHNwAN3VEesXqrF5ctSXu9rKDjkSCXIJ5OekWZNciYgLQQR8PlruNdnhfywYfxSF2Zhk5PsrMQibUI2JhtCg2x+AjQSHDt8×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×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×5PC3W8VTc/384Spbt91LUpjyfnbSB8p8EdQLXEOFTEIZXdv7R9lxo+UWHUaQuSESykJXLQyEqSkIWVDOopdalfiITnLX2hmkYoiakolz1zEkaTE5bHcAjT339VLHVdksJzZBqFgG5GxF7EPpbpqYfFd6iZarYmwKoAQpKpQe3fUC4B/KOXhHRuGMQShEw08tSwE3WWKlqSbDYMxPWEGWrtZQUt0kJ7qQRlmN+IpKn3bu22jGEVyynJLQFIF1m9sxZidByA98NJNttCKqSOhcRk19HmSDLmBz2ZLv4EgPpq2×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×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×8JSKelmMSe82oGu0UcKkTJkxAEtYlSQQJl8pId1E7kn4RBT0qVDMUAHMpzzH9IRwhrFVyM8k1OTvgF4bQsxWNLub36xcn4jdgWAjTEpgSDp198BVTuusN2VhairlpW6VEh/q/1pDdhFAMSmBUyaQUBgm3n9COfU9S2/wBeMHMExYyZ6FJ3LEc30+UH2Q6vR8NyZIZ3bnFkyJOhSIohVQoPlAB0ciIjh1QovnSnwc/pE0j3Q28vsuzpckfgDxhNZKAbKn3RWOBqPtzleQb4mN5WAIG61ef6CCoL0gPJJ9yZujEUA6CN51cnUM0T0vD8l3MsqHUq/WJzgKNpQA6/zMMDlg9GLpZnEYGJJ5iCX9hAfgR6CJ5WCjkjXkP0iKxWaGUHPjGpSIkWLmNMpirUssrYhUTJcpSpSilXT9DaOQD7TMX/AGgyO2BOfKxlo56+yNrx2ObIKg2xgBM4KklfaMM3OClQLI0rxRaAoz0×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×8+TdiT0GEYHh4LhJJN7qVfyBiCsrKdLgAJSnlCNmmCyiv66vGUys+gUX3OnpFZbRZrK/MpWTSLVHTtJWOaFX8QREdJhtw+0XsTmdlTzV8kkAczt7zAbvoaKo5PxHg5pakyge7lRMQS10zEpUPFnbyh04f4TRUpQuVOKXCc0lCAZg1dipYGTXvAEszgbw/bFR9lXSU7ikkv4pK0W/hiphxCpYTvtdiDqG87ahrefTcm4pnIS5Y38TcR11HTBMqmRKpUNLaaAqasXZSk91gWNm8es1MJVZTEyky01CUFQQgsVsHKcj3UbsUxzPHBPKsswkgPucvKwJtfaB9IFO2bzPzg62rBfoNYlhtat1ikqcg1PYTWtq5ywG7WwgthWJLlTE5VqQXspCin1ys490NieG5GJFUxNSJNSLzR2YIWAPbyAp77aqSSCzsCSSVKuwUIlPM+v5xiZOOdBB0UnTWx0hoxXhekpkgrxNJUfwinU/i3aO2m3rC7OwpSJ9OlwtM1aMkxDlCwVAEpJALjQpIChuBBTTAfUMikQEIcP3Q3pu0TIko/KInCDlA5ARqpMM2REa0bAD0jTJd4laMNAuh9UzIjwMYePQBqoyrSKqFa+MWTFiVTpbSHi0iuasGmckbRtLmJO0Zn0aEi5v1inLAGhisIUGQBy0UJpBLiNTeMhAgkMARNLIuSwAuTt4xGvIlJUoskXJMIvG/FTyFy5QKAvuk/iIGvgP0MK3RErB+N8QmurBLlH7qW9ue5zHq20DsVqgh3U7agabtz09bdYBS640tEtaCRMnKKQTZkJuW8T4WG8VFrVM7FIVdaQSTs9/hf+l6qvkcMyp5U5dkjmT5WJLaMf6MHqq1c1WVDqJ7oAFy+gAGp5b6xbrR93lStMuSmypq3GYixSgBLzFa91O+pF4FzatMtBSgKlJI7yyQJ80HUWcSUEWKQSSzFRgxiRs9+6UQFhdQCQ4Lyqc2YuLTJ45B0pPMjuluGgJQmzhcISSFG5MyY6QTzPtF7XSNYVqZSpqkpQkBLsGGnLb+ekNnFCP2WmTTgjOSFzAG1YhKeeno/N4aX0CL9idiExU6ZkT3lqUEpY3KlFm8zHXPtMRlSikl2RLSinQBoBlGY/wAAbzjm3A1J2uKUSSbduhXjkOc+uVo6txhKK64Ol2zKI5ZhLAbr3Yq8l0kW+Ktpsr8DUQpy7ahj7oOYrTF8w0v9eMa0VN3bbxalzj7KtowXfZ0aFKppS728/wCUayKTkG6c/rnB+tp3MRypECwlSVT5Rfx/rFBUlVTVU9MkWMxK19ESyFKfoWy+KhFzFagaDeLnDtRKoaOoxKddV0ID65PZQOqplvIcosxR2kJmnrA5v9rGIdvi07lKCZIb+6HPnmUr0ijSqIYMWF7N15/GAParmLVMWXUtRWpR3KyST6wxUiwUsRoAzf0/nHSkqSRyY8sixz7xAULHla+2w+jC9QqyzQBaGirlgo0Z/c+31/VSz/e9QWPrrDQ5TQJdhGtcEF2izJq1JAWCyhoQoght9XiCsSFAH6+EU1XR7oKVoHRbxbiKfUpCZh7RvZUQkqYf3mze+DnBfFqqdBlhFtznUFAjcEukfw/rASkQlKbi/wDS3OKkghMw/R+ngtJqkQ6nJ+1SelViSn8q2Ud92c6bHeG7/wDKlNLpBOnJOcnLll3B3Bd7DXnoY4Ash9ngvhNXLCSienOhVlDfxBcOoWN/hC1QUzreHfaiZqTM7FARsHLnz/lBCR9qVE7TcyOoIUPPQxyKjUJMjKiyVEqN7sdBccm6XgGjCv2iYQCEHd9B5wF32NZ9QYZjdLU3kT5cw/lB738JvF9Qj5xwrApVOoEiYuY4Y9pkQ+lslyfEt4QwYt9ocyVlkS5iyRu5J/iU6m8SYOxDs65sWJU8tpHDZXHVQlQeaog8yCPfpDjhvGEwodUtKjzBKfdDbRFaY1zS5Lm7xoBES3c+Mby0EloUhKmI6+slyEGZOWEpHPU+A3hdx3jKVImdhKaZN3a4T0tqfq8cz+0PG5sxQkknPMIB21Zh4QNvSDQ18QcWKqVU6EpKJS1BYFyVJ2JIs+jDqfGFLFKztVLOgTYB9g/xPxiTFKnLiCUo0lshOgskBIB6Ns+haFisqFCaq+p9/nvCJNsZtIs8Q1QCJaEn2B/N9YtUeKS5FPLmMFz1JCUBYdCAnK6yPxFywTp3S7vCzVrJMShbyQNdr8gXt4n4RbpwhNuQjMxEqV2q5qZkzZS1Pl6JSWSEjkBEEuTnXmWoLP8AiT81QOpEPmDaJJ9IpqhlAFj1Q1lPSI7ZZzTNUy0LS7tZyl2AOpN+QgTPxJVSVzlICEp5Em/K/iNekAainysCO9v57QbwKnVMkmXLTmmKmME7lwlgOd/fCuCSsO18Bz7KZWfF6csfu8yj/CR846xxAWxFT3CkI/1RzX7GJCjiaSNkEl/y3BYc7NHQuKS2IqbRMpHh+LTpeMnk9Gnxf2/0GqcXDb/N/mPhGainCrixiKlBygvb5X+UWJa7NvGWjeUKqSXHuiBSmQSByghVTOY0+cVVqCv8P62HzhGhhfkYYuompQkZQTdR/CNSfIfGAX2v12WWmmSMssKyy0DYJuVK6nMSddUvd46JUThLlqWklJUyE5SkHXve0CwskP1jjXFqDPnqU47NCciFH+Jcw+Kn8gOcavHhXJi8mbbaQtSVBEtzfT3P+oixTYshx3iPEAj0tFbHq6WpRRJSUSxoCSS3UnUneBBEdBQtWzDdDzJmicCAUnTTQOLEgsQIVsSpyib13Hhy6Mx84pyJ6kEFJYj69IKV1Z26AojvpuTzHTf9GiKOrI3aJKeozJy7i7RVIuobO8R0q+cWJywL7nf6+ESqZDVKydDGgXqTGiVsI1mqsxiUAzTqJufOJzMJsDbQxUTMYWjaWvKMx1OkFogTXWlDaWs2×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×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

Dunno…

Can you hum a few bars?

It was a young Martin Shaw. It was aimed at me…

We are sitting in different rooms.

buffy said:

We are sitting in different rooms.

not talking ):

buffy said:

You need to do someone I might know…

F’r‘instance…

I second that. Someone guessed that my “Afghan Girl” was Tom Baker.

I tried Mr Bean, and failed to get a small file. Mr Bean ended up looking like Salvador Dali as painted by Picasso.

499 bytes.

I’m not sure if this is cheating … I’ve made the outer parts of the pace lower-res (more blocky) than the crucial parts (the eyes and nose).

dv said:

499 bytes.

I’m not sure if this is cheating … I’ve made the outer parts of the pace lower-res (more blocky) than the crucial parts (the eyes and nose).

That’s a tough one.

dv said:

499 bytes.

I’m not sure if this is cheating … I’ve made the outer parts of the pace lower-res (more blocky) than the crucial parts (the eyes and nose).

Neutron Starr?

The Invisible Man?

Aaand again

dv said:

499 bytes.

I’m not sure if this is cheating … I’ve made the outer parts of the pace lower-res (more blocky) than the crucial parts (the eyes and nose).

kryten said:

dv said:

499 bytes.

I’m not sure if this is cheating … I’ve made the outer parts of the pace lower-res (more blocky) than the crucial parts (the eyes and nose).

Neutron Starr?

dv said:

Aaand again

dv said:

499 bytes.

I’m not sure if this is cheating … I’ve made the outer parts of the pace lower-res (more blocky) than the crucial parts (the eyes and nose).

I don’t know. It’s humanoid, probably male.

Stephen Colbert ?

dv said:

Aaand again

dv said:

499 bytes.

I’m not sure if this is cheating … I’ve made the outer parts of the pace lower-res (more blocky) than the crucial parts (the eyes and nose).

John Hewson.

Aaand again

dv said:

499 bytes.

I’m not sure if this is cheating … I’ve made the outer parts of the pace lower-res (more blocky) than the crucial parts (the eyes and nose).

Mandibular Prognathism

dv said:

Aaand again

dv said:

499 bytes.

I’m not sure if this is cheating … I’ve made the outer parts of the pace lower-res (more blocky) than the crucial parts (the eyes and nose).

Possibly Gregory Peck.

Trying the same thing on the Weaving picture I can get the byte count down by 498 but it really does become a bit less recognisable. That peripheral information seems somewhat crucial after all.

Looks a bit like Ewan McGregor…

furious said:

Looks a bit like Ewan McGregor…

One of his countrymen

Robert the Bruce.

dv said:

furious said:

Looks a bit like Ewan McGregor…

One of his countrymen

Christopher Edelstein?

Phil_C said:

dv said:

furious said:

Looks a bit like Ewan McGregor…

One of his countrymen

Christopher Edelstein?

Make that Eccleston.

dv said:

furious said:

Looks a bit like Ewan McGregor…

One of his countrymen

Billy Connoly

dv said:

Trying the same thing on the Weaving picture I can get the byte count down by 498 but it really does become a bit less recognisable. That peripheral information seems somewhat crucial after all.

First one and second one
.

I disagree. I find that the second one is more recognisable. There’s a light patch on his forehead, not a dark patch.

mollwollfumble said:

I disagree. I find that the second one is more recognisable. There’s a light patch on his forehead, not a dark patch.

Fair enough.

Here’s one with just the central facial features, for 499 bytes.

dv said:

mollwollfumble said:

I disagree. I find that the second one is more recognisable. There’s a light patch on his forehead, not a dark patch.

Fair enough.

Here’s one with just the central facial features, for 499 bytes.

that’s the lady from that mulligrub tv show..

Urgh, make it stop :(

• that’s the lady from that mulligrub tv show..

I was thinking the joker but that works too…

furious said:

• that’s the lady from that mulligrub tv show..

I was thinking the joker but that works too…

dv said:

mollwollfumble said:

I disagree. I find that the second one is more recognisable. There’s a light patch on his forehead, not a dark patch.

Fair enough.

Here’s one with just the central facial features, for 499 bytes.

Adding the hair, chin and ear increases the byte count to 688, but gives much better context.

dv said:

mollwollfumble said:

I disagree. I find that the second one is more recognisable. There’s a light patch on his forehead, not a dark patch.

Fair enough.

Here’s one with just the central facial features, for 499 bytes.

Liza with a zee?

dv said:

dv said:

mollwollfumble said:

I disagree. I find that the second one is more recognisable. There’s a light patch on his forehead, not a dark patch.

Fair enough.

Here’s one with just the central facial features, for 499 bytes.

Adding the hair, chin and ear increases the byte count to 688, but gives much better context.

Doesn’t look like Liza now.

Looks familiar, but I’m not sure who.

Anne Hathaway?

furious said:

Anne Hathaway?

Doing a google image search on the second one gets some fairly amusing results.

Computers are still bloody hopeless at recognising stuff, unless there is an exactly matching image on line.

dv said:

dv said:

mollwollfumble said:

I disagree. I find that the second one is more recognisable. There’s a light patch on his forehead, not a dark patch.

Fair enough.

Here’s one with just the central facial features, for 499 bytes.

Adding the hair, chin and ear increases the byte count to 688, but gives much better context.

Michael Jackson?

furious said:

Anne Hathaway?

Correct.

But mollwolfumble has set a very tough challenge, it is hard as nails to get a monochrome png of a real face recognisable at 500 bytes or less. That seems to be a very hard limit, at least using my methodology. Perhaps some very clever curve and polygon fitting svg file could do better.

All very interesting.

furious said:

Anne Hathaway?

Yes, very good

kii said:

Urgh, make it stop :(

Very well.

dv said:

furious said:

Anne Hathaway?

Yes, very good

Doesn’t look anything like her.

https://en.wikipedia.org/wiki/File:AnneHathaway_CUL_Page4DetailB.jpg

The Rev Dodgson said:

dv said:

furious said:

Anne Hathaway?

Yes, very good

Doesn’t look anything like her.

https://en.wikipedia.org/wiki/File:AnneHathaway_CUL_Page4DetailB.jpg

Hathaway is believed to have grown up in Shottery, a village just to the west of Stratford-upon-Avon, Warwickshire, England. She is assumed to have grown up in the farmhouse that was the Hathaway family home, which is located at Shottery and is now a major tourist attraction for the village. Her father, Richard Hathaway, was a yeoman farmer.

The Rev Dodgson said:

dv said:

furious said:

Anne Hathaway?

Yes, very good

Doesn’t look anything like her.

https://en.wikipedia.org/wiki/File:AnneHathaway_CUL_Page4DetailB.jpg

Really showing your age there… what are you about 400yo?

The Rev Dodgson said:

dv said:

furious said:

Anne Hathaway?

Yes, very good

Doesn’t look anything like her.

https://en.wikipedia.org/wiki/File:AnneHathaway_CUL_Page4DetailB.jpg

Very drole m’lud

• Richard Hathaway, was a yeoman farmer.

Need good soil to grow yeoman…

I’m not making any comment about whether I knew of the other Anne Hathaway.

Witty Rejoinder said:

The Rev Dodgson said:

dv said:

Yes, very good

Doesn’t look anything like her.

https://en.wikipedia.org/wiki/File:AnneHathaway_CUL_Page4DetailB.jpg

Really showing your age there… what are you about 400yo?

I’ll admit that I don’t have any clue what either of the Hathaway ladies looks like.

sibeen said:

Witty Rejoinder said:

The Rev Dodgson said:

Doesn’t look anything like her.

https://en.wikipedia.org/wiki/File:AnneHathaway_CUL_Page4DetailB.jpg

Really showing your age there… what are you about 400yo?

I’ll admit that I don’t have any clue what either of the Hathaway ladies looks like.

do you have the internet whereon you can do a google image search. by any chance?

Bogsnorkler said:

sibeen said:

Witty Rejoinder said:

Really showing your age there… what are you about 400yo?

I’ll admit that I don’t have any clue what either of the Hathaway ladies looks like.

do you have the internet whereon you can do a google image search. by any chance?

I’ve given it up for lent.

sibeen said:

Witty Rejoinder said:

The Rev Dodgson said:

Doesn’t look anything like her.

https://en.wikipedia.org/wiki/File:AnneHathaway_CUL_Page4DetailB.jpg

Really showing your age there… what are you about 400yo?

I’ll admit that I don’t have any clue what either of the Hathaway ladies looks like.

I probably could not have picked Will’s country wife out of a line up.

You must have seen something with Hathaway. She has been in all kinds of movies.

And why didn’t she change her name if they were really married?

Peak Warming Man said:

And why didn’t she change her name if they were really married?

Because his name is a bit silly

I mean you must have seen The Dark Knight Rises OR Brokeback Mountain OR Les Miserables OR Interstellar, right?

dv said:

I mean you must have seen The Dark Knight Rises OR Brokeback Mountain OR Les Miserables OR Interstellar, right?

Inexplicable though it is, some people are just not into popular culture.

dv said:

I mean you must have seen The Dark Knight Rises OR Brokeback Mountain OR Les Miserables OR Interstellar, right?

Err…no.

Really.

dv said:

I mean you must have seen The Dark Knight Rises OR Brokeback Mountain OR Les Miserables OR Interstellar, right?

I’ve seen Les Mis, but I’m pretty sure it didn’t have either of the Hathaway chicks in it.

sibeen said:

dv said:

I mean you must have seen The Dark Knight Rises OR Brokeback Mountain OR Les Miserables OR Interstellar, right?

Err…no.

Really.

+1.

sibeen said:

dv said:

I mean you must have seen The Dark Knight Rises OR Brokeback Mountain OR Les Miserables OR Interstellar, right?

Err…no.

Really.

Back away slowly, he’s gone full Bubblecar.

Witty Rejoinder said:

dv said:

I mean you must have seen The Dark Knight Rises OR Brokeback Mountain OR Les Miserables OR Interstellar, right?

Inexplicable though it is, some people are just not into popular culture.

I’m pretty much into unpopular culture these days.

The Rev Dodgson said:

dv said:

I mean you must have seen The Dark Knight Rises OR Brokeback Mountain OR Les Miserables OR Interstellar, right?

I’ve seen Les Mis, but I’m pretty sure it didn’t have either of the Hathaway chicks in it.

Let me guess, you saw some live theatre production of it in 1982 or something, right?

The Rev Dodgson said:

dv said:

I mean you must have seen The Dark Knight Rises OR Brokeback Mountain OR Les Miserables OR Interstellar, right?

I’ve seen Les Mis, but I’m pretty sure it didn’t have either of the Hathaway chicks in it.

Actually The Rev’s response has made me review my history and I’m afraid I must admit that I have seen Les Mis. The year was 2000 and the town was somewhere in west end London. I still cannot recall Ms Hathaway.

dv said:

The Rev Dodgson said:

dv said:

I mean you must have seen The Dark Knight Rises OR Brokeback Mountain OR Les Miserables OR Interstellar, right?

I’ve seen Les Mis, but I’m pretty sure it didn’t have either of the Hathaway chicks in it.

Let me guess, you saw some live theatre production of it in 1982 or something, right?

Don’t be silly.

It was more like 1995.

Looking at her IMDb listing I don’t think I have seen any of her movies, yet I know who she is…

Bubblecarism is expanding rapidly. Pretty soon no one will have heard of anyone.

• Bubblecarism is expanding rapidly. Pretty soon no one will have heard of anyone.

Who is this Bubblecar then?

• Bubblecarism is expanding rapidly. Pretty soon no one will have heard of anyone.

Who is this Bubblecar then?

furious said:

• Bubblecarism is expanding rapidly. Pretty soon no one will have heard of anyone.

Who is this Bubblecar then?

Heh.

furious said:

• Bubblecarism is expanding rapidly. Pretty soon no one will have heard of anyone.

Who is this Bubblecar then?

Someone who used to post here I think.

But who are you?

• But who are you?

Hang on, I will check IMDb…

furious said:

• But who are you?

Hang on, I will check IMDb…

yep, that sounds about right…

Furious

furious said:

• Bubblecarism is expanding rapidly. Pretty soon no one will have heard of anyone.

Who is this Bubblecar then?

Name- Bubblecar
Born-: Yes.
Address-: Classified
Occupation-: Epicurean, Artist.
Religion-: No.
Bowling Average-: Very.
Batting Average-: Yes.
Achievements-: Being defrocked, cashiered and drummed out of the comments section of every major newspaper in the world.

I think this is recognisable (with squint) at 450 bytes.

Method.
Decrease colour depth to 3.
Decrease size by 4 to 64 × 64.
Save as 20% jpeg greyscale.
Display in forum at 256 × 256

dv said:

Bubblecarism is expanding rapidly. Pretty soon no one will have heard of anyone.

I hadn’t heard of Hugo Weaving.

Is that his occupation?

I think this is recognisable (with squint) at 450 bytes.

Method.
Decrease colour depth to 3.
Decrease size by 4.
Save as 20% jpeg (greyscale if appropriate)
Display in forum 4 times larger.

Same here with the others, except 30% jpeg on Afghan girl because definitely ugly at 20% jpeg.

Celebrity A 480 bytes and B 430 bytes

Celebrity C 390 bytes and D 440 bytes

What I find definitely weird is that face recognisability as a function of number of bytes isn’t all that dependent on image size, and whether it’s a cartoon or photograph. And only depends a small amount on familiarity of image.

Here are the detailed steps involved in the byte reduction process.
Final sizes left to right 1180 bytes, 470 bytes, 470 bytes

Original image

Cartoonise to 3 colours

Reduce size by 4

jpeg compress to 20%

I could even claim from this that at 470 bytes, Cat from Red Dwarf is more recognisable than the Mona Lisa.

mollwollfumble said:

I think this is recognisable (with squint) at 450 bytes.

Method.
Decrease colour depth to 3.
Decrease size by 4.
Save as 20% jpeg (greyscale if appropriate)
Display in forum 4 times larger.

Same here with the others, except 30% jpeg on Afghan girl because definitely ugly at 20% jpeg.

Celebrity A 480 bytes and B 430 bytes

Celebrity C 390 bytes and D 440 bytes

What I find definitely weird is that face recognisability as a function of number of bytes isn’t all that dependent on image size, and whether it’s a cartoon or photograph. And only depends a small amount on familiarity of image.

We should collab on a paper.

mollwollfumble said:

dv said:

Bubblecarism is expanding rapidly. Pretty soon no one will have heard of anyone.

I hadn’t heard of Hugo Weaving.

Is that his occupation?

He is a famous actor, known for

The Adventures of Priscilla, Queen of the Desert
The Matrix (and its sequels)
The Lord Of The Rings trilogy
Captain America
Cloud Atlas
The Hobbit trilogy

and so on

dv said:

He is a famous actor, known for

The Adventures of Priscilla, Queen of the Desert
The Matrix (and its sequels)
The Lord Of The Rings trilogy
Captain America
Cloud Atlas
The Hobbit trilogy

and so on

Fairly well known since the ‘Bodyline’ mini-series in the 1980s.

Bubblecar would suck at my current uni unit of Screen Celebrity and Stardom. It’s about how a celebrity is made i.e. made a commodity and the old Hollywood star system.

dv said:

mollwollfumble said:

I think this is recognisable (with squint) at 450 bytes.

Method.
Decrease colour depth to 3.
Decrease size by 4.
Save as 20% jpeg (greyscale if appropriate)
Display in forum 4 times larger.

Same here with the others, except 30% jpeg on Afghan girl because definitely ugly at 20% jpeg.

Celebrity A 480 bytes and B 430 bytes

Celebrity C 390 bytes and D 440 bytes

What I find definitely weird is that face recognisability as a function of number of bytes isn’t all that dependent on image size, and whether it’s a cartoon or photograph. And only depends a small amount on familiarity of image.

We should collab on a paper.

Not a bad idea. On the topic of minimum number of bytes needed for facial recognition.

Completely unrelated search lead me to find:

Perception of Pixelated Images

Review

dv said:

We should collab on a paper.

Also thanks to furious, will need those references.

First draft of paper collab.

Minimum number of bytes needed for facial recognition

Darren Voss (aka dv) and David A Paterson (aka mollwollfumble)

Abstract

A sequence of facial recognition experiments was put before a panel of ? interested individuals of mixed age and sex in an internet environment. The first two experiments were of minimum pixel resolution with squared pixels needed for recognition of the face of a Hollywood celebrity, cartoon, famous painting, iconic stencil, famous photograph, and well known comedian. The cartoon and ? required the fewest pixels for recognition () and the Hollywood celebrity the most pixels (). But pixels aren’t the same as bytes. The next two experiments looked at facial recognition using cartoons drown from faces, and separately using strong jpeg compression. In all three cases, it was possible to get reliable facial recognition at between 600 and 650 bytes but impossible in less than 500 bytes. In the final experiment, cartoon-like colour compression was followed in turn by pixel reduction and jpeg compression. With this strategy it was possible to get good facial recognition in every case with between 400 and 500 bytes and in one case easily recognisable with 400 bytes. One startling conclusion is that image size and image subject had no noticeable effect on the number of bytes needed for facial recognition, and that previous familiarity with the image had only a very small effect on the number of bytes.

Introduction.

Facial recognition software is big business . It is now used in recognising people from their passport photographs , and used by the police . Pixilation and blurring are frequently used to conceal identities . There have been several scientific studies on how the pixel resolution of square pixilation affects facial recognition . The topic has even appeared in a comedy show .

Experiment 1.

Experiment 1 started with an unfamiliar greyscale 256 × 256 photograph of the rather unremarkable face of a Hollywood celebrity, chosen to be a name that wouldn’t be guessed at random. Pixelated versions of this with square pixels at increasing resolution were placed before an internet panel of interested people. This panel consisted of ? people of mixed sex, predominantly of white Anglo-Saxon origin, of ages 40 to 80.

The participants were only told that it was an image of an individual, ruling out at the start pictures of buildings and countryside. They were not told whether the picture was of a human, or whether there was any cartoonish distortion. The square pixels were presented at resolutions of ? x ?, ? x ?, ? x ?, 11 × 11, 13 × 13, 16 × 16, 24 × 24 and full resolution

The panel discerned that the picture was of a face at a resolution ? x ?, and a human face facing front left at ? x ?. At a resolution of 13 × 13 it became possible to distinguish for the first time between the light-coloured face oval itself and the light parts of the far background and vee neck. At this resolution it was clearly a male face with no significant face hair, receding hairline, mid-brown hair looking towards but to the left of the photographer. This resolution was also the first time that the cartoon possibility could be eliminated.

At 16 × 16 there was no further progress in identifying the individual. At 24 × 24 the individual was correctly identified as Hugo Weaving, although two panel participants thought wrongly that it looked more like Kevin Kostner.

Participants had been squinting to take the corners off the sharp edges of the pixellation, and as an extra test the 11 × 11 image was enlarged for viewing using a smoothing algorithm that removed all the sharp edges. The recognisability of the face did not improve.

There were several elements in the pixellated images that were observed to have a big effect on recognisability, and on later images as well. At coarse resolution the location of the nose was absent and this had a strong negative effect on recognition. At finer resolution the shadow of the nose connected to the mouth and this also had a strong negative effect on recognition.

Experiment 2.

Experiment 2 looked at square pixellation of a wider range of faces that were thought would be easier to recognise, but again were chosen to be unlikely to be guessed at random. Experiment 1 had chosen a Hollywood celebrity. The faces chosen for Experiment 2 were those of (A) a famous painting – the Mona Lisa, (B) a cartoon – Mickey Mouse, © a famous photograph – the Afghan Girl, (D) a photograph that is familiar as an iconic stencil – Che Guevara, and (E) a comedian – Marty Feldman. The first three were in colour and the last two in greyscale.

Initial images were smaller than the one in Experiment 1, and only approximately square. Participants on the internet panel were told only that they had to identify a face. The Mona Lisa was unrecognisable at ? x ? and identified at ? x ?. Mickey Mouse was unrecognisable at ? x ? and identified at ? x ?. The Afghan girl was unrecognisable at ? x ? and identified at ? x ?. Che Guevara was unrecognisable at ? x ? and identified at ? x ?. Marty Feldman was unrecognisable at ? x ? and recognisable at ? x ?.

When we say “recognisable” for the first two experiments, that doesn’t mean that all panel participants were able to identify all images, only that at least one participant made a correct identification. One participant had all the identifications wrong in this experiment at the same resolution that another participant had all the identifications correct. The colouring had a big influence on the recognisability of both the Mona Lisa and the Afghan Girl. Without colour, they could only have been recognised at higher resolution. Let us call this “bare recognisability”. In subsequent experiments a different standard was used, in which everyone would agree that the face was recognisable. Let us call this “easily recognisable”.

As with Experiment 1, sporadic testing using a smoothing algorithm that removed sharp edges prior to viewing did not improve recognisability.

Before Experiment 2, it was expected that Mickey Mouse would be recognised at lower resolution than any other face, in the experiment it tied for first place with the Mona Lisa. In an experiment reported in Scientific American in ?, the most recognisable pixellated face that they tried was the Mona Lisa, with George Washington being the second most recognisable.

Experiment 3.

Attention now turned to the minimum number of bytes necessary for facial recognition. There is a world of difference between pixel resolution and number of bytes. The 3 × 3 jpeg image of Mickey Mouse after expanding to 151 × 153 for viewing used 1519 bytes whereas the 16 × 16 jpeg image of Hugo Weaving after expanding to 256 × 256 for viewing used only 983 bytes. Clearly the jpeg compression algorithm was having some effect. Preliminary tests using images from Experiments 1 and 2 showed that recognition is sometimes possible at little more than 600 bytes.

It was thought that scalable vector graphics might improve the byte count, but available svg images of Che Guevara on the web require an embarrassingly large 28.5 kB or more.

On the other hand, the very first identikit from 1959 contained 500 elements . If these are considered to be divided among 10 or more facial features (face shape, eyes, nose, ears, mouth, hair, hair length, etc.) then the selection of elements requires 150 or fewer bits, 19 or fewer bytes. This would have to be added to the number of bytes required for each selected facial feature.

With this in mind, face recognition was tried with black and white cartoons.

Which cartoons? How created? What recognisability? How many pixels? How many bytes?

Experiment 4.

Another way to reduce the byte count is to use extreme jpeg compression. The image viewing and processing program IrfanView allows jpeg compressions from 1% (most compression) to 100% (no compression). There is no direct translation of these numbers into the number of bytes in the compressed file.

Applying 1% jpeg compression, with greyscale conversion where appropriate, directly to all the original images yielded recognisable images of the following sizes.
Hugo Weaving – 646 bytes
Mona Lisa – 583 bytes
Mickey Mouse – 684 bytes
Afghan Girl – 543 bytes
Che Guevara – 575 bytes
Marty Feldman – 614 bytes
These sizes are startlingly consistent between face images of many different origins, subjects, colour choices and number of pixels.

An image that has been subjected to extreme compression in this way has a colour range reduced to 16 colours, square pixellation at a resolution of ? x ? (or equivalent for rectangular images ), with sub-pixel variation in some of the squares. As the percentage increases, the first change is that the amount and delicacy of the sub-pixel variation improves. This sub-pixel variation easily picks out and adds the eyes, mouth, ears and hairline, but doesn’t give them the correct shape.

More variations were carried out, combining size reduction with varying levels of jpeg compression. Some variations, typically greater size reduction with a larger jpeg percentage, improved image quality with the same number of bytes, but not enough to have a great effect on the minimum number of bytes needed for facial recognition. (see image).

Experiment 5.

In this final experiment, the techniques from Experiments 3 and 4 were combined to improve on the seeming limit of 500 bytes. The final version selected is as follows. Experiment 3 tried cartoons, and the first step in Experiment 5 is to turn the image into a cartoon by reducing the number of colours in the image to three. Experiment 4 tried jpeg compression and size reduction, and in Experiment 5 the cartoonisation is followed by a 4-fold reduction in size followed by a jpeg compression to 20%. Only the Afghan Girl was not identifiable, so the final jpeg compression percentage there was increased to 30%.

This experiment yielded easily recognisable images of the following sizes.
Hugo Weaving – 456 bytes
Mona Lisa – 479 bytes
Mickey Mouse – 487 bytes
Afghan Girl – 438 bytes
Che Guevara – 400 bytes
Marty Feldman – 447 bytes

Conclusions

These sizes in bytes are startlingly consistent between face images of many different origins, subjects, colour choices and number of pixels. Hugo Weaving, specifically selected to have a face that is difficult to identify, and placed in a large initial image, requires very nearly the same number of bytes for easy recognition as Mickey Mouse, specifically selected to have a face that is easy to identify.

It appears as if every face that has been seen before by participants often enough for recognition, can be easily recognised in an image requiring fewer than 500 bytes, and sometimes as little as 400 bytes.

Awaiting feedback from dv.

Fair format.

I would use photographs of somewhat famous people, people you’d expect the audience to recognise in nice crisp photographs, but not ultra-mega-famous iconically famous.

I would also, for the proper paper, drop the Mickey Mouse and Mona Lisa, and stick with real filtered photographs of real humans.

And I’d avoid famous photographs (So I wouldn’t use my Hitchcock one, or the Che one, or “Afghan Girl” Gula.)

And we’d need about 100 test subjects.

• And we’d need about 100 test subjects.

You would also need them to identify the clear pictures at the end to ensure they actually know who they are…

furious said:

• And we’d need about 100 test subjects.

You would also need them to identify the clear pictures at the end to ensure they actually know who they are…

And make sure no one on this forum is involved because somehow this little bubble of the internet was cut adrift from Western Civilisation sometime in the 19th century.

dv said:

Fair format.

I would use photographs of somewhat famous people, people you’d expect the audience to recognise in nice crisp photographs, but not ultra-mega-famous iconically famous.

I would also, for the proper paper, drop the Mickey Mouse and Mona Lisa, and stick with real filtered photographs of real humans.

And I’d avoid famous photographs (So I wouldn’t use my Hitchcock one, or the Che one, or “Afghan Girl” Gula.)

And we’d need about 100 test subjects.

I don’t have about 100 test subjects handy right now.

I think we probably already have enough results from the thread already without collecting more, but feel free to round up more people, I can keep supplying crushed images if you want.

furious said:

• And we’d need about 100 test subjects.

You would also need them to identify the clear pictures at the end to ensure they actually know who they are…

Good point, Furious. Imagine having 100 Bubblecars as your test subjects.

• And make sure no one on this forum is involved because somehow this little bubble of the internet was cut adrift from Western Civilisation sometime in the 19th century.

I recently floated across on a raft so I got a couple of your pics…

Do a survey monkey titled can you name these famous people? And spam fb with the link…

poikilotherm said:

Do a survey monkey titled can you name these famous people? And spam fb with the link…

You wont believe what these five famous people have been up too.

• Do a survey monkey titled can you name these famous people? And spam fb with the link…

You won’t believe what these celebrities look like…

• You wont believe what these five famous people have been up too.

Damn you!

I don’t have about 100 test subjects handy right now.

I think we probably already have enough results from the thread already without collecting more, but feel free to round up more people, I can keep supplying crushed images if you want.

We could find them on farcebook.

You could just say.. We bet you can’t guess who these 100 famous people are? And then say like and share..

sarahs mum said:

I don’t have about 100 test subjects handy right now.

I think we probably already have enough results from the thread already without collecting more, but feel free to round up more people, I can keep supplying crushed images if you want.

We could find them on farcebook.

You could just say.. We bet you can’t guess who these 100 famous people are? And then say like and share..

too late she cried as waved her wooden leg

You could probably get a studio audience after they’ve been poled.

dv said:

Fair format.

I would use photographs of somewhat famous people, people you’d expect the audience to recognise in nice crisp photographs, but not ultra-mega-famous iconically famous.

I would also, for the proper paper, drop the Mickey Mouse and Mona Lisa, and stick with real filtered photographs of real humans.

And I’d avoid famous photographs (So I wouldn’t use my Hitchcock one, or the Che one, or “Afghan Girl” Gula.)

And we’d need about 100 test subjects.

I see whole point of the paper as being the startling result that famous faces are not significantly more recognisable (bytewise speaking) than faces that are not ultra-mega-famous iconically famous. Therefore it becomes necessary to include the ultra-mega-famous iconically famous faces, like Mickey and Mona.

Also the secondary thrust of the paper is the almost equally startling result that it doesn’t matter bytewise speaking if the source is a painting, photograph, stencil, or cartoon. Therefore essential to include Che, Afghan girl etc. And it doesn’t matter whether the actor is a serious actor or comedian.

Totally counterintuitive, therefore publishable.

Ditto colour vs black and white.

If you only put in greyscale Weaving-like faces then you lose the whole purpose of the paper, it wouldn’t prove anything, or add anything valuable to the sum total of human knowledge.

like Mickey and Mona
—-

use Queen liz and Washington.

sarahs mum said:

like Mickey and Mona
—-

use Queen liz and Washington.

We haven’t all seen the Mona Lisa or shook hands with Mickey. We know them from the amount of times they have been replicated and appropriated.

Same goes for the people on the money.

Well, who are you? I really wanna know. Tell me, who are you? ‘Cause I really wanna know.

ChrispenEvan said:

Well, who are you? I really wanna know. Tell me, who are you? ‘Cause I really wanna know.

Damn, that’s an ear worm, off to YouTube to scratch it.

ChrispenEvan said:

Well, who are you? I really wanna know. Tell me, who are you? ‘Cause I really wanna know.

Well I can tell you that out here in the fields I fight for my meals

sarahs mum said:

like Mickey and Mona
—-

use Queen liz and Washington.

Famous leader and famous leader? Bit of duplication there if it isn’t done right.

Looking at the pictures of George Washington on the Web, I don’t even recognise it in the original. Only an American would.

Why not add them in as well? We want examples from the entire widest possible range of face images.

I was thinking of Munch “The scream”. And the face of a fox.

Queen Liz.

George Washington

A fox

mollwollfumble said:

sarahs mum said:

like Mickey and Mona
—-

use Queen liz and Washington.

Famous leader and famous leader? Bit of duplication there if it isn’t done right.

Looking at the pictures of George Washington on the Web, I don’t even recognise it in the original. Only an American would.

Why not add them in as well? We want examples from the entire widest possible range of face images.

I was thinking of Munch “The scream”. And the face of a fox.

Queen Liz.

I’m not American but can recognise George Washington. On the othre hand, except for the fact that she was on a stamp I would not have recognised a yound Lizzy.

sibeen said:

mollwollfumble said:

sarahs mum said:

like Mickey and Mona
—-

use Queen liz and Washington.

Famous leader and famous leader? Bit of duplication there if it isn’t done right.

Looking at the pictures of George Washington on the Web, I don’t even recognise it in the original. Only an American would.

Why not add them in as well? We want examples from the entire widest possible range of face images.

I was thinking of Munch “The scream”. And the face of a fox.

Queen Liz.

I’m not American but can recognise George Washington. On the othre hand, except for the fact that she was on a stamp I would not have recognised a yound Lizzy.

Well you might not recognise a young GW

sibeen said:

mollwollfumble said:

sarahs mum said:

like Mickey and Mona
—-

use Queen liz and Washington.

Famous leader and famous leader? Bit of duplication there if it isn’t done right.

Looking at the pictures of George Washington on the Web, I don’t even recognise it in the original. Only an American would.

Why not add them in as well? We want examples from the entire widest possible range of face images.

I was thinking of Munch “The scream”. And the face of a fox.

Queen Liz.

I’m not American but can recognise George Washington. On the othre hand, except for the fact that she was on a stamp I would not have recognised a yound Lizzy.

Lizzy’s nose isn’t really that long is it?

mollwollfumble said:

sibeen said:

mollwollfumble said:

Famous leader and famous leader? Bit of duplication there if it isn’t done right.

Looking at the pictures of George Washington on the Web, I don’t even recognise it in the original. Only an American would.

Why not add them in as well? We want examples from the entire widest possible range of face images.

I was thinking of Munch “The scream”. And the face of a fox.

Queen Liz.

I’m not American but can recognise George Washington. On the othre hand, except for the fact that she was on a stamp I would not have recognised a yound Lizzy.

Lizzy’s nose isn’t really that long is it?

(Tongue in cheek warning). We can’t have an older Lizzy because she resembles Dolores Umbridge.

Just joking.

This was a good one, but it’s an etched image like George.

dv said:

sibeen said:

I’m not American but can recognise George Washington. On the othre hand, except for the fact that she was on a stamp I would not have recognised a yound Lizzy.

Well you might not recognise a young GW

Not a chance in the world that I would recognise this chap.

mollwollfumble said:

dv said:

sibeen said:

I’m not American but can recognise George Washington. On the othre hand, except for the fact that she was on a stamp I would not have recognised a yound Lizzy.

Well you might not recognise a young GW

Not a chance in the world that I would recognise this chap.

Bloody troublemaker by the looks of him.

mollwollfumble said:

This was a good one, but it’s an etched image like George.

She was wattley when I was growing up.

mollwolfumble’s comments on svg made me look into. It seems like a lot of the svg files going around are full of “cruft”, legacy shit that aren’t actually required.

And, at 498 bytes, who is this?

dv said:

And, at 498 bytes, who is this?

gene wilder?

sarahs mum said:

dv said:

And, at 498 bytes, who is this?

gene wilder?

YES!

dv said:

And, at 498 bytes, who is this?

King Catactacus?

kii said:

dv said:

And, at 498 bytes, who is this?

King Catactacus?

or King Caractacus….aaargh, it’s Rolf Harris!

kii said:

kii said:

dv said:

And, at 498 bytes, who is this?

King Catactacus?

or King Caractacus….aaargh, it’s Rolf Harris!

I think you should check your cataracticus

dv said:

kii said:

kii said:

King Catactacus?

or King Caractacus….aaargh, it’s Rolf Harris!

I think you should check your cataracticus

It’s been done twice this year. Thanks.

kii said:

dv said:

kii said:

or King Caractacus….aaargh, it’s Rolf Harris!

I think you should check your cataracticus

It’s been done twice this year. Thanks.

No need to thank me. I live only to serve others.