Hacker News new | past | comments | ask | show | jobs | submit login
Paint.wtf (paint.wtf)
191 points by hgarg on Jan 19, 2021 | hide | past | favorite | 82 comments



Got to 4 or 5 on a few of them with NO. Then discovered a solid color put me on #1.

AI is only as smart as it’s programming (which can be pretty darned smart, but this AI wasn’t trained to deal with people who are trying to break it).

On a side note, Roboflow is freaking amazing.

https://paint.wtf/ranking/5qms/pORkuKYlFhXWjn6SWD4M


And now it's ruined as so many submissions are just a solid color. It's really too bad, because if you go through the submissions there are some really amazing ones in there.. particularly for raccoon driving a tractor.


Yeah it's a shame, I get that someone tries to beat the ai, but you would think that it's less fun to draw the eleventh solid color when the top ten is already filled with them.

On a more positive note: There are some awesome Giraffe in the arctic drawings as well. Like this masterpiece https://paint.wtf/ranking/IdF9/vBRMWbTGG9Td6oeMoim8

This is really fun to do, I'm currently ranked 36th for the racoon drawing and 15th for the giraffe (So top 5 if you ignore the solid colors), but I should probably do some work right now.


This giraffe is awesome. I wonder why it is ranked lower than this one: https://paint.wtf/ranking/IdF9/Y8CmWvlELvy1MoSU8Iv3


Solid colors get higher ranks


An example of humans breaking an AI's utility function, rather than the other way around!


Omg I thought the app was just broken and not loading drawings! I guess technically it's still broken, but just not in the way I thought.


Oh that's why, I thought that was a bug with Firefox.


Honestly, I thought it was broken and failing to load the images until I read this.


I think it says a lot about the state of the web and web development that this was the first conclusion people drew.


This is an interesting observation. The cool part about CLIP is that we can ask it any prompt we want. So “fixing” this loophole is just a matter of prompt engineering.

We already added a penalty to the score for text by subtracting similarity to “handwritten text”.

Any ideas what these solid color blocks would rank highly for that and actual drawing wouldn’t? I can update the prompts and regenerate the leaderboards this morning.


Check my comments here; top with simple negativity -- but straight lines. Simple solid also works for some reason.


Hey everyone! We made this as a hackathon project over the weekend to experiment with OpenAI’s new CLIP model. https://openai.com/blog/clip/

Glad you’re enjoying playing with it and discovering failure modes! Yesterday we tweaked the prompts to penalize people writing text when we discovered CLIP can read. Looks like we should add something to prevent it from falling in love with solid blocks of color. Any ideas for how to describe that? (Someone below suggested a bonus for “complexity” but I’m concerned that’d make it fall in love with scribbles.)

You can read more about the project on my Show HN from yesterday: https://news.ycombinator.com/item?id=25822610


Update: "An illustration of an empty canvas filled with color" seems to work pretty well to extract the solid blocks of colors from the actual drawings.

We're adding that to our prompts now and then will recalculate the leaderboards. Please surface any other failure modes you find! It's pretty neat to learn about what trips it up; it's brand new so you all are amongst the first to poke at it.


Update 2: appears this was actually a bug in my code that applies a text penalty :blush:

These blocks of color weren't scoring highly for the prompt... they were scoring really low for similarity to text... which ended up giving them a sort of anti-text bonus relative to other submissions. The fix should be up shortly.

Update 3: leaderboard should be mostly fixed!

Update 4: leaderboard fix fully implemented. And they're looking good! There are some artists amongst you. https://paint.wtf/leaderboard


I drew a circle and it said "Potentially NSFW". Neat idea, thanks for sharing.


You can click to reveal those if you want.

It runs each image through a variety of prompts including screens for racist symbols, nsfw drawings, and nudity. If you send me a link to yours I can let you know what it tripped up.

(One of our heuristics is "is it closer to nudity than to the prompt" so if it wasn't a good representation of the prompt it could be more that it was far away from the target vs being particularly close to the nsfw target.)


A bonus for complexity can be given according to something like this:

https://ai.stanford.edu/blog/topologylayer/


Giraffe in the arctic, after seeing other submissions I think no other is similar: https://paint.wtf/ranking/IdF9/mHjesZ4qkfj6JqljXsdk


Looks like I had the same idea, although I admit yours looks a bit better :)

https://paint.wtf/ranking/IdF9/XTLr1H02fW1vegFtR7rb


I sort of love that the AI prefers solid gray over any sort of creativity or complexity. A preview of... gray goo!

https://en.wikipedia.org/wiki/Gray_goo


> I sort of love that the AI prefers solid gray over any sort of creativity or complexity

Well that's this specific AI...

> A preview of... gray goo!

We are the gray goo, or rather: our ecosystem is. A flood of self-replicating machines that took over the planet a long time ago, eating everything. If we introduced a new kind of nanite-based life that actually managed to devour our ecosystem (which is probably more challenging than it sounds, given that there are real thermal limits on how fast such molecular disassemblers can work), I'm "optimistic" that they would differentiate into another complex ecosystem over time as well.


AI prefers 1-color rectangle to anything else :)

Reminds me of the time I added neural networks to my asteroids-like game (with gravity and walls). I recorded myself playing the game, added sensors to detect walls, enemies, powrups around the player's ship and the output was any combination of "throttle, rotate left, rotate right, shoot".

The AI analyzed hours of my recorded games and decided the best strategy is to constantly hold the throttle and do nothing else :)


My perfect upside down dinosaur got me #1234, which is pretty cool, but also...it's clearly #1

https://paint.wtf/ranking/5qms/7gRBc8y3ArSbKDXnr5a7


I did this for the monster. #1, which I think is pretty decent given the GUI.

https://paint.wtf/ranking/6RWC/qesvGIv33OP443peodpB


How do you like my selfbalancing bulldozer?

https://paint.wtf/ranking/ZJ7I/5t4EkiSzZICiubjwK9EZ

Unfortunately not as pretty as a solid gray :)


Its funny how everyone who really tried seems to be going for the carrots :)


Carrot is aparently the first vegetable that comes to most people's mind.[0][1]

[0] - http://sciencechatforum.com/viewtopic.php?p=264070

[1] - https://www.bbc.com/future/article/20160504-the-two-word-gam...


Unfortunately I suspect the bot mostly appreciated the solid black background.


Well, it doesn't have the artistic purity of the magenta square, but it's definitely spooky.


You have encountered a Grue.


Hmmm...

https://paint.wtf/ranking/7sAX/drZSMbNMv1avmID4M2XS

2 out of 291 for a solid green square. I feel like there might be a "complexity" metric missing in the model.


The top submission is pretty awesome though https://paint.wtf/ranking/7sAX/1xScOGJT8gIOQ2u09bBK

That being said, looking at pretty much every other top-ranking drawing shows there's definitely a need for some sort of "low quality" check


Complexity is a cool thing to ask it about and adjust the score based on. I’m concerned it’d start loving scribbles though.

Any other ideas for ways we could help it understand what we’re asking of it?


106th for exploding pig :( I thought I did rather well https://imgur.com/a/SVbO2K3


Perfect, missing only a speech bubble with "NANI?!"


No appreciation for real art by this AI, it seems.


It's already ruined. All this shows to me is that the best way to train an AI using people is to not tell the people they're training an AI.

Edit: I just imagined that it would post submissions to reddit instead and have people vote up and down. No mention of AI, just voting up and down on other people's art.

Then that could train the AI, right? I have no idea, never worked with ML.


The cool thing about this model is that it actually doesn’t need to be trained! The way to tweak it is just to frame your question in the right way.

We’re doing to update the prompts probably to penalize it for similarity to “a solid block of color” or something like that.


Voting may not produce desired training data either ... Painty McPaintface ...


I was confused why my "Tattoo of a mailbox" was flagged as NSFW until I looked closer:

https://paint.wtf/ranking/Y08Q/II8fc1RgEJXcXk2f4vhG


thanks for the laugh! haha I love this


https://paint.wtf/ranking/Y08Q/u1JcNVbWAjG6e2RkDIID

1 out of 197 for a tattoo of a mailbox

Skynet still a long way away


Of course perfection only comes with persistent practice, my natural talent was not enough to project me to such a hallowed height of artistic accomplishment.

I had to go through https://paint.wtf/ranking/Y08Q/wRC0z2rQmSYrcuBFeNh2 (no. 4) and https://paint.wtf/ranking/Y08Q/YXQduzfjXldDduZa0FYK (no. 6) first.



The AI tends to prefer solid blocks of color over actual drawings, and a particular shade of pink over other colors. However, I think this is the case because the drawings are largely terrible. For one of the challenges that I saw, I couldn't beat a very well-drawn giraffe with a solid color. Of course, it's possible that for that particular challenge a different color from the standard pink would have been more highly rated.


This is a really creative use of what I assume is CLIP. Seems like there are really simple hacks though, like choosing the right solid background color


Just me or the submissions can’t be viewed in Firefox for iOS, even with tracking protection turned off?


look at top comment, he found the best way to get perfect score is a solid color for the whole image


I also cannot view the submissions on PC


You can, they're just solid gray (that's what this particular AI thinks is the most scary monster or whatever else it's evaluating)


This made me so happy. I get high rank for just drawing green on green background. And I always thought my drawing skills sucked :)

Joke aside, this is a cool idea! With some tweaking for not identifying pure colored images as high rank it would be more fun for the artists among us!


Made this before realizing that blocks of colors were the way to go. Feel sad now.

https://paint.wtf/ranking/6RWC/HKJvfYwu0zxVWTF7zIB7


Nice! We're recalculating all the leaderboards so no worries; check back in a couple hours to see your new ranking.


My "World's most fabulous monster" https://paint.wtf/ranking/6RWC/5Ml7o0ENsU91WobWIJtE



Nice, is your drawing the green upside down long neck dinosaur?

I drew almost the same thing except I didn’t color mine only drew the outline and mine was facing the other way and also it was smaller, probably because I drew it on my phone. Guessing that you drew on a laptop or desktop, or that you zoomed out if zooming out is possible.

I got 97th place at the time of my submission, and now I am currently 108th place out of 388 total.


I got 1 for making a purple square with decreased saturation. Purple works better than the other colors for some reason, it would be really interesting to see the training data set.


No, I did only the outline also, in black, with one eye. Currently 11 out of 604 submissions. I used the mouse on a desktop computer and used 2 strokes to do the whole thing.


4/610

I didn't colour it either, actually.

https://paint.wtf/ranking/5qms/KoDe48hxCVFtila50VBu


Best I've got was 7 / 44 - for the Racoon on Tractor


I see your 7/44 and I raise you a 4/56.

https://paint.wtf/ranking/WPhs/N0NCdWh5snaUFrlXOhLF


lol!


Proud of my $5 toilet [1] being #38 out of 250+. It represents a dog bowl...

Btw, didn't work for me in Firefox.

[1] https://paint.wtf/ranking/SGlSX/011pQIJbBSlV3ZZ1dANJ


I drew a circle and it got flagged nsfw


Circles are kind of hot though...


Looks like color filling shapes is better ranked than strokes drawings. My artwork : https://paint.wtf/ranking/ZJ7I/3UdyfyDLMYLomWO1wKkF


I drew a "bumblebee that loves capitalism" staring out from his tower window over a busy city of hives, surrounded by pots of honey: https://paint.wtf/ranking/yBUM/rfPayDhXUYz09y0RfG5k

I was placed at #1491 directly below someone who drew one individual black dot[1], and above someone who drew a single squiggly line[2].

[1] https://paint.wtf/ranking/yBUM/GQvMN5TVvPiSkqepGiLl

[2] https://paint.wtf/ranking/yBUM/4Dz9vh9O3P2ZAYx5uyY6


I had a look at the scoring. It looks like it penalized you pretty heavily for all the "H" letters on the honey.

We added a text penalty because it turns out the model can read and so all the top results were people writing the prompt. Not sure how to tell the model to only penalize bad text but let Ok text through.


Interesting, I had suspected I was penalised because the actual bee is such a small part of the image, and facing away from the viewer.


In the leaderboard, the images are too small. It would be nice to be able to see them without clicking on them individually.


Presumably this site is dead now - it just showed "Showing Results" when I tried it (after drawing a picture).


You must have caught us in the middle of the leaderboard update job. If you go back via the homepage it should link directly to your submission for that prompt so you can see your rank.


I am drawing arbitrary things, poorly, and haven't placed worse than 250. Amusing in a way.


I was so proud of my upside down dinosaur.

Rejected.


Looks like there was a lot of Rothko in the training set


The neat thing is: there is no training set! The CLIP model[1] takes arbitrary text prompts. So for these we're just sending eg "An illustration of an upside down dinosaur" and it returns a similarity value between the image's encoding and the text's encoding.

To filter out things we don't want (eg text and nudity) we penalize the score based on the similarity for eg "A nsfw illustration".

[1] https://openai.com/blog/clip/


For me the bucket fill didn't work


It works fine - you just have to change the bg colour to a solid colour instead of a pattern.


doh, thanks!


i wonder how it would ever tell the difference between broccoli and a mushroom cloud


Terrible app. If you paint solid blocks of color it ranks the picture extremely highly. Actual good art was thrown to the back of the pile.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: