Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

The Generative Burrito Test

generativist.com

Post reply on HN
Print view View original

51–56 of 56 posts

Re: The Generative Burrito Test

#51

Post by AlotOfReading »

autoexec wrote: post #48

I don't eat a lot of burritos and when I do they aren't bean burritos, so I'm honestly wondering: do they commonly have whole beans in them? I expect that if they do, they aren't often so clean and shiny looking, but what I expected is more of a mushy/refried bean look. Do people get burritos with beans in them more or less as pictured? Aesthetically, it seems like it'd look pretty appealing if you were someone who l…

It's entirely possible to have clean, whole beans in a burrito. It's unusual in commercial kitchens because whole beans are kept warm in the cooking broth until service to avoid drying out. The preparer will scoop them out with a slotted spoon to drain, but usually they're in too much of a hurry to fully drain it. Product photos don't rush this step because a soggy burrito doesn't look good on camera and they also undercook things so the ingredients don't mush. AI tools have ingested a lot more product photos than real burritos.

Re: The Generative Burrito Test

#52

Post by beefnugs »

recursivecaveat wrote: post #50
kbenson wrote: post #8

Oh wow, I've been hearing about Nano Banana Pro in random stuff lately, but as a layman the difference is stark. It's the only one that actually looks like a partially eaten burrito at all to me. The others all look like staged marketing fake food, if I'm being generous (only a few actually approach that, most just look wrong).

I don't know if it's the abundance of stock photos in the set or the training, but the 'hypertune' default look of AI photos drives me crazy. Things are super smooth, the colors pop wildly, the depth of field is really shallow, everything is overly posed, details far too sharp, etc. Vaguely reminds me of the weird skin-crawler filter levels used by people like mr beast. I think it is the fine tuning, because you can…

This is what we deserve for not burning to the ground every company with fake food ads as soon as it started

Re: The Generative Burrito Test

#53

Post by zenoprax »

Aloisius wrote: post #42
zenoprax wrote: post #31

Earlier quoted context omitted.

Hyperrealistic beans? The focal plane? You are reaching really hard here. The table grain is the only thing that gives it away - if it weren't for that no one without advance warning is going to notice that it's not real.

Yeah, hyperrealistic beans. They don't look real at all. The inside of an actual burrito is messy after you bite into it (and usually before). That burrito has a couple of nearly dry, yet for some reason speckled, beans that look more like they're floating on top of the burrito rather than actually in it. And yeah, the focal plane is wonky. If you try to draw a box around what's in focus, you end up with something th…

If you're approaching it from a "semantic pixel peeping" perspective then yes, I understand what you mean. It's a pretty clean bite... but it's important to remember the context in which most images will be assessed.

Earlier this week I did some A/B testing with AV1 and HEVC video encoding. For similar bit rates there was a difference but I had to know what to look for and needed to rapidly cycle between a single frame from both files and even then... barely. The difference disappeared when I hit play and that's after knowing what to look for.

For anyone curious: if you are targeting 5-10 Mbps from a Bluray source AV1 will end up slightly smaller (5-10%) with slightly more retention of film grain in darker areas. Target 10 Mbps with a generous buffer (25 MB) and a max bit rate (25 Mbps) and you'll get really efficient bit rates in dark scenes and build up a reserve of bandwidth for confetti-like situations. The future is bright for hardware video encoding/decoding with royalty-free codecs. Conclusion: prefer AV1 for 5-10 Mbps but it's no big deal if it's not an option.

Re: The Generative Burrito Test

#55

Post by fruitworks »

beefnugs wrote: post #52
recursivecaveat wrote: post #50

Earlier quoted context omitted.

I don't know if it's the abundance of stock photos in the set or the training, but the 'hypertune' default look of AI photos drives me crazy. Things are super smooth, the colors pop wildly, the depth of field is really shallow, everything is overly posed, details far too sharp, etc. Vaguely reminds me of the weird skin-crawler filter levels used by people like mr beast. I think it is the fine tuning, because you can…

This is what we deserve for not burning to the ground every company with fake food ads as soon as it started

https://www.youtube.com/watch?v=Vy5vkTqpk8M

Re: The Generative Burrito Test

#56

Post by qingcharles »

BoorishBears wrote: post #34
minimaxir wrote: post #22

One of my tests for new image generation models is professional food photography, particularly in cases where the food has constraints, such as "a peanut butter and jelly sandwich in the shape of a Rubik’s cube" (blog post from 2022 for DALL-E 2: https://minimaxir.com/2022/07/food-photography-ai/ ) For some reason ever since DALL-E 2, all food models seem to generate obviously fake food and/or misinterpret the fun co…

I tried having Claude generate a prompt for Seedream and got this: https://imgur.com/a/6xX5TDE I can kind of see what you mean in that it went for realism in the aesthetics, but not the object... but that last one would probably fool me if I was scrolling

Here's Nano Banana Pro, which nailed it, IMO, but I had to fudge the prompt:

https://imgur.com/8kMqbBO

"peanut butter, jelly and bread rubik's cube. each smaller cube in the rubik's cube is one ingredient, randomly selected. professional food photography style. ensure it looks like a working rubik's cube"

Post reply on HN