How AI Image Generators Work (Stable Diffusion / Dall-E) - Computerphile

แชร์
ฝัง
  • เผยแพร่เมื่อ 3 ต.ค. 2022
  • AI image generators are massive, but how are they creating such interesting images? Dr Mike Pound explains what's going on.
    Thumbnail image partly created by DALL-E with the prompt: "Computerphile TH-cam Video presenter Mike Pound Explains Diffusion AI methods thumbnail with green computer style title text on a black background with grey binary"
    / computerphile
    / computer_phile
    This video was filmed and edited by Sean Riley.
    Computer Science at the University of Nottingham: bit.ly/nottscomputer
    Computerphile is a sister project to Brady Haran's Numberphile. More at www.bradyharan.com

ความคิดเห็น • 756

  • @InfinityDz
    @InfinityDz ปีที่แล้ว +763

    Glad to have finally found someone I can actually listen to about AI, someone that doesn't hype things up and isn't trying to sell me something.

    • @Isaac-wr8et
      @Isaac-wr8et ปีที่แล้ว +17

      If it needs other people's intellectual property to work, then it is a legal concern for those whose work it being exploited without consent.
      This is why Getty images is suing, and why Adobe is building their AI Firefly off of licensed work, and making guidelines to compensate those who stock are being used by their AI.

    • @theguyshadows
      @theguyshadows ปีที่แล้ว +10

      @@Isaac-wr8et Why do ignorant people like you comment? This is literally an explanation of how these types of AI work, no different than explaining other AI models. Nothing about this is abstract mumbo jumbo BS, you just don't understand it, lmao.

    • @youssefabusamra3142
      @youssefabusamra3142 ปีที่แล้ว +60

      @@Isaac-wr8et salty credit obsessed artist spotted

    • @Isaac-wr8et
      @Isaac-wr8et ปีที่แล้ว +16

      @@youssefabusamra3142 the way I see it, people aren't happy when their private information or content is downloaded/stolen without permission by the government or AI. The source that these AIs are using too "learn", is doing just that.

    • @youssefabusamra3142
      @youssefabusamra3142 ปีที่แล้ว +53

      @@Isaac-wr8et "the government or AI" my brother in christ they uploaded these photos for public recognition agreeing to the site's terms and conditions
      what the AI is doing is the equivalent of "looking" at these photos and recognizing features and patterns. If you're so dead set on the stealing narrative then pick a real painting that was used for training and try to "steal" it by recreating it with prompts

  • @ayushdhar
    @ayushdhar ปีที่แล้ว +477

    A deep dive on the google colab code would be amazing!

    • @talestomazalves8089
      @talestomazalves8089 ปีที่แล้ว +3

      ☝☝

    • @nevokrien95
      @nevokrien95 ปีที่แล้ว +1

      There r videos that do that.
      Also there r alot of papers explaining the system in detail.

    • @rayankhan12
      @rayankhan12 ปีที่แล้ว +6

      @@nevokrien95 Can you provide links to those videos please?

    • @nevokrien95
      @nevokrien95 ปีที่แล้ว +2

      @@rayankhan12 sure i will send it in parts here is some code in pytorch (i personally know only tensorflow but i still got the gist of how i would go about doing it)

    • @nevokrien95
      @nevokrien95 ปีที่แล้ว +3

      @@rayankhan12 this is abit mathematicly involved. Key point it shows is that the limiting behivior is not what makes these work. They r simply autoencoders with some extra whistles

  • @Qman621
    @Qman621 ปีที่แล้ว +933

    Stable diffusion doesn't actually actually apply noise to images, it uses a compressed low dimensional latent representation of the image and applies noise to that. The model is running in this abstract latent space, and then the autoencoder recreates the image afterwards.

    • @michaelpound9891
      @michaelpound9891 ปีที่แล้ว +375

      Great point. Yes I skipped over this mainly for the sake of the length of the video. This also explains the slightly odd brown noise we see in the video, which is actually a low noise latent passed back through the VAE decoder.

    • @asdf30111
      @asdf30111 ปีที่แล้ว +54

      @@michaelpound9891 if you tell AI to run for zero steps you can look at the noise.

    • @threeMetreJim
      @threeMetreJim ปีที่แล้ว +14

      I once used an overtrained network to store several images, then manipulated the low dimensional part to give some trippy image tweening (a few inputs reperesenting this is image 1,2,3,... etc. and then getting the in-between images). Unfortunately very low resolution, and took ages, guess a web-browser isn't the place for running neural nets...

    • @JordanMetroidManiac
      @JordanMetroidManiac ปีที่แล้ว +43

      @@threeMetreJim That’s always a fascinating experiment with VAEs. Encode two items to two latent points, take the midpoint of the two latent points, and then decode that latent midpoint to see what the resulting item is. I tried this with music and it was interesting to hear a transition from Beethoven to Schubert.

    • @xlspiorp
      @xlspiorp ปีที่แล้ว

      fatter and older

  • @wlockuz4467
    @wlockuz4467 ปีที่แล้ว +642

    Finally! Ever since Stable Diffusion was released I was looking for an explainer on how it worked that wasn't "Oh it generates images from noise" or something that went too deep into technicals that I didn't understand.
    Very beautifully explained Dr. Mike Pound! Hope you do another video where you dive into the code where we can see the parts which were visualized here.
    One thing that's still unclear to me is how was the network trained to relate text with images and how does it utilize this information when actually producing images?

    • @generichuman_
      @generichuman_ ปีที่แล้ว +50

      @@thebirdhasbeencharged I don't understand why people answer questions they don't know the answer to. He's asking how the diffusion model which starts from purely random noise, uses the text embedding generated from clip to guide the diffusion. "A.I. is just fancy pattern matching" is about as unhelpful an answer as you could imagine.

    • @IceMetalPunk
      @IceMetalPunk ปีที่แล้ว +27

      I'm a little unclear on that myself, but the best understanding I can manage is that the CLIP (language model) embeddings must be included in the diffusion network's training. So while it's learning how to predict the noise on a picture of, say, a bunny, it's also given the text description of the bunny, which means it's learning how the descriptions affect the noise at the same time as it's learning how the underlying picture does.
      I think. As I said, not 100% clear on that, so don't take my word for it 😅

    • @dialecticalmonist3405
      @dialecticalmonist3405 ปีที่แล้ว +24

      They asked humans, "Is this a frog? Yes or no."
      They took that data to develop an AI that could be asked, "Is this a frog? Yes or no."
      They did the same with "stilts".
      They did the same with "on". The difference being that they used a variety of known "objects" to determine whether they were "on" something or not.
      They also probably classified "on" as a verb, rather than a noun. This makes it a union of two objects. A union associated with "proximity" or something like that.
      Like he said at the end, they need an intact frog and intact stilts as a requirement in the "frog on stlits" image. So they look like "frog feet with proximity to stilt objects" etc.
      I would assume human objects on stilts strongly guided their classification of frog objects on stilts.

    • @IceMetalPunk
      @IceMetalPunk ปีที่แล้ว +17

      @@dialecticalmonist3405 I guess that's a decent high-level explanation, but I would clarify that it's actually not using any kind of classification system. Classifiers are an entirely different family of neural networks. The guidance system used here is a transformer-based language model, which is less like asking "is this a frog (y/n)?" and more like asking "here's an image, describe what it is".

    • @ScumfuckMcDoucheface
      @ScumfuckMcDoucheface ปีที่แล้ว +7

      @@generichuman_ haha I'm assuming they deleted their unhelpful message, as I don't see it. Great call out. =)

  • @kgsz
    @kgsz ปีที่แล้ว +3

    Came here by accident and man, aren't you the gifted one? I was engrossed in the video knowing barely anything about the technologies and techniques uses, and I don't feel dumber -- that's an achievement :)
    Thanks again, will pop here often.

  • @carlborgen
    @carlborgen ปีที่แล้ว +144

    Would have been nice hear a bit more about the "gpt-style transformer embedding". Wouldn't those classifications have to be included in the training data already?

    • @yiding
      @yiding ปีที่แล้ว +23

      This is basically what CLIP does. CLIP learns from a massive amount of image-description pairs using GPT-style (Transformers) encoding so that it can map texts and images. CLIP data are not classification labels. Then the difference between the texts and the generated images can be calculated and minimized.

    • @hermezkonrad
      @hermezkonrad ปีที่แล้ว +3

      Key word is embeddings. Initial feature space of text has two bad properties: it has big dimensionality (each token is it's own dimension essentially) and sparsity. By using Transformers you compress representation of this object in more compact and dense form, so it's easier to work with.

  • @myce-liam
    @myce-liam ปีที่แล้ว +178

    Pounding that like button! You guys have inspired me to start an undergraduate degree in Cyber Security - thank you for all of your videos!

    • @emmafountain2059
      @emmafountain2059 ปีที่แล้ว +18

      I watched Dr Mike Pound's video on Convolutional Neural Networks when it first came out and it got me into machine learning. Now I'm doing undergrad computer vision research with CNNs. It's honestly kind of crazy to think about how much this channel has affected my life.

    • @myce-liam
      @myce-liam ปีที่แล้ว

      @@emmafountain2059 Well played! Are you enjoying yourself doing the computer vision research?

    • @themariokartlick
      @themariokartlick ปีที่แล้ว +1

      @@emmafountain2059 yup. I started watching this channel when I was a bored IT audit intern who hated doing work papers. I literally sat in the bathroom or went on walks outside and just watched. Now I’m a pentester. The reach of high quality TH-cam channels like Computerphile are hard to measure but I don’t think I’m unique.

    • @arvindhmani06
      @arvindhmani06 ปีที่แล้ว

      @@emmafountain2059 That's amazing! Wish you the best!

  • @aijeveryday_guy
    @aijeveryday_guy 9 หลายเดือนก่อน +6

    I couldn't agree more! Since the release of Stable Diffusion, I've been searching for an explanation that strikes the right balance between simplicity and technicality. Your video did an excellent job of providing a clear understanding without overwhelming us with excessive technical details. Dr. Mike Pound, you have a remarkable talent for explaining complex topics in a beautifully straightforward manner!

  • @serhat757
    @serhat757 ปีที่แล้ว +5

    Can't believe Mike can effortlessly make that shape with his hand (little finger) at 5:37

  • @Zothaqqua
    @Zothaqqua ปีที่แล้ว +31

    I tried to guess how these things work. Now I'm taking the difference between my guess and this explanation and feeding it to my neurons. Thanks!

  • @rickhobson3211
    @rickhobson3211 หลายเดือนก่อน +1

    I followed some of that.. but some of that also sounded a lot like Michelangelo's "start with the block of marble and carve away everything that doesnt look like "X." I will come back to watch this again after the first watching settles! Thank you for providing this.

  • @dileepvr
    @dileepvr ปีที่แล้ว +103

    12:58 I'd like to hear more about that GPT-style transformer embedding of text. Was text part of the training set?

    • @tristanstevens6162
      @tristanstevens6162 ปีที่แล้ว +31

      yes they used image-text pairs dataset (LAION) to train the guided diffusion model

    • @ekki1993
      @ekki1993 ปีที่แล้ว +4

      @@tristanstevens6162 That dataset accounts for understanding text too? Like, if you have sets of images of cats and sets of images of frogs and very few sets of different animals being "fused" (e.g. maybe one capybara that looks like a dog), how would the neural network get to the interpretation of what a cat-frog means as an image? Is LAION that big? Or does the GPT neural network somehow bridge that gap?

    • @DontThinkSo11
      @DontThinkSo11 ปีที่แล้ว +27

      @@ekki1993 a lot of the job of "understanding" is being done by the embedding network, which was trained on a very large corpus of words. So while the training set for stable diffusion might not have any examples of a frog-bunny fusion, CLIP is able to take the phrase "frog-bunny fusion" and transform it into a vector that encodes something about the meaning of the phrase. Stable diffusion was trained conditioned on this embedding, so it generally has learned to take concepts from this embedding and include them in the image. The hope is that stable diffusion is able to generalize across all concepts that can be represented by the embedding, so that even if it hasn't seen this specific thing before, it has seen similar stuff and is able to still produce a reasonable image that matches the concepts in the embedding.

    • @ekki1993
      @ekki1993 ปีที่แล้ว +1

      @@DontThinkSo11 Thanks for the answer!

    • @mvbmvb
      @mvbmvb ปีที่แล้ว

      @@ekki1993 It's possible for all/any of the components to contribute to the result working - like, even if it hasn't seen any pictures by an artist x, "style of artist x" may still work as a prompt if it's seen text describing them.
      This is an issue for artists who've been complaining that image generators can reproduce their style in some ways. It means that nothing can be done to prevent this; a base model might still understand them if the images aren't in the set. "Worse", fine-tuning seems to work well enough that you can add in new concepts and styles at home even if they're not in the model originally.

  • @danieletorrigiani
    @danieletorrigiani ปีที่แล้ว +5

    Wow! Had not seen listing paper since my dad was trying to teach me basic on a commodore 64. Had no idea it was still a thing. Big jump from having to read code on paper to make sense of it to this.

  • @froop2393
    @froop2393 ปีที่แล้ว +5

    would like to see more details but the explanation was superb for an introduction

  • @juliankandlhofer7553
    @juliankandlhofer7553 ปีที่แล้ว +1

    Oh i DEFINITELY want to see mike's deep dive into the code!

  • @mtalsi1
    @mtalsi1 5 หลายเดือนก่อน +3

    I love that he's doing all of this on 1980s printer paper. Proper geek

  • @chaoslab
    @chaoslab ปีที่แล้ว +1

    Been listening to house music in the background (on the low down) when the odd watching computerphile / numberphile for quite a while now.
    Thought it was time to fess up.
    Vibing it is probably just me on this tip.

  • @beachdancer
    @beachdancer 10 หลายเดือนก่อน +4

    the explanation sounds like magic. It is like a sculptor saying he just chips away pieces of the stone until he finds the horse hidden inside.

    • @grafzeppelin4069
      @grafzeppelin4069 19 ชั่วโมงที่ผ่านมา

      It's strikingly similar, except a sculptor starts with a goal image in mind, but AI image generation doesn't; it just has general "knowledge" of "associations" between the words of the prompt and parts of images.

  • @lpbaybee4942
    @lpbaybee4942 ปีที่แล้ว +1

    The best compsci content on the internet, period.

  • @tobiasjennerjahn8659
    @tobiasjennerjahn8659 ปีที่แล้ว +5

    Great explanation. Just complicated enough to understand for someone who keeps up with this stuff on the surface level, but isn't interested in reading the papers. Thanks.

  • @rajeshprajapati1851
    @rajeshprajapati1851 ปีที่แล้ว +4

    Dr Mike Pound with pen and paper can make me understand any topic.

  • @Thermophobe
    @Thermophobe ปีที่แล้ว +4

    i am still a bit confused about the process but loved the ending!

  • @Jinjukei
    @Jinjukei ปีที่แล้ว +21

    Thank you so much for talking about this topic! Great and very enjoyable!

  • @francoisjohannson139
    @francoisjohannson139 ปีที่แล้ว +6

    As i understand it, it's somehow like an subtractive synthesizer in audio,the A.I. has the role of the filter.

  • @cyndi5hunt
    @cyndi5hunt 11 หลายเดือนก่อน +5

    I love this methodical and interesting presentation of the process. It is inspiring as always! 💜

  • @Ultimatro
    @Ultimatro ปีที่แล้ว +47

    So stable diffusion is just the AI version of that sculpting joke: Start with a big block and take away the parts that dont fit

    • @IceMetalPunk
      @IceMetalPunk ปีที่แล้ว +13

      Yep! It's a bit like apophenia, like looking at random clouds and seeing coherent shapes in them, but with some priming about what you "should" be seeing :)

    • @housellama
      @housellama ปีที่แล้ว +18

      "I saw the angel in the marble and carved until I set him free. ”
      - Michelangelo

    • @morpheus6749
      @morpheus6749 ปีที่แล้ว

      No, that's not how it works at all. His explanation is highly inaccurate and misleading, which is throwing you off. Try reading the actual papers on the subject, or going through the code.

    • @morpheus6749
      @morpheus6749 ปีที่แล้ว

      No, that's not how it works at all. His explanation is highly inaccurate and misleading, which is throwing you off. Try reading the actual papers on the subject, or going through the code.

  • @jmalmsten
    @jmalmsten ปีที่แล้ว +11

    I remember rewatching Brows Held High's episode on the movie Blue. In it, there are clips from the film. It was just a blank blue screen. On film. Sot there was some noise from the grain. It was from a DVD rip (I think). And it had therefore, by the time it got from the film to my computer screen on youtube, gone through numerous re-encodings. Encodings that expect visual interests and details to compress.. but those had none. So I noticed that the artifacting was picked up as not-noise. And it tried to encode it as if it was normal video. And through the generations of transfers, the blue blank screen was now... Filled with random shapes of blue tones that had gotten enhanced over time.
    I joked then that we were basically seeing the encoders hallucinations. Little did I know, that a few years later, seceral image processors would spring up that essentially used that method, but guided. And they would be able to hallucinate pretty high resolution images... From noise ..

  • @kevinb3300
    @kevinb3300 ปีที่แล้ว +2

    Awesome video. That’s the clearest explanation I’ve seen. The hand drawn explanation explained it so perfectly. Would love to see a follow up video that goes through the code. Also would be awesome to include examples when talking about the muppets in the kitchen, etc.

  • @Mutual_Information
    @Mutual_Information ปีที่แล้ว +280

    Add noise to images and train a model to undo that addition.. then you have something that maps from noise to images.
    One thing I find so impressive about these researchers.. is that they would try this. It’s so bizarre.. just because, from a distance, it’s not at all clear that such a task is doable.

    • @Alex-ye8qp
      @Alex-ye8qp ปีที่แล้ว +21

      And with the intention to guess how much noise a picture would have only by text description as an input haha

    • @rickyspanish4792
      @rickyspanish4792 ปีที่แล้ว +40

      Right, it sounds like a completely non-intuitive way of going about it, and yet, that's what ended up working. They must have iterated on a gazillion different ideas before they landed on this one.

    • @MrHaggyy
      @MrHaggyy ปีที่แล้ว +19

      The idea of adding something, do a transformation and substract to get only the transformation of the data or the something is actually quite common in math and control theory.
      Real hit from AI was to get "the" transformation from data, something and output into an algorithm. This general function of transformation is what allows the image generation. We give it data or something that is slightly off, and amplify the error by the transformation.
      😅 it's some weird combination of the chicken egg paradox and a rock paper scissor but with data, something, algorithm and output.

    • @ekki1993
      @ekki1993 ปีที่แล้ว +30

      It's why science works better by being public and not subject to short/mid-term revenue. There were already teams training models to undo noise, there was already GPT to interpret text, there was already a database of millions of text-to-image pairings and there were already models trying to feed text to image-based neural networks. Kinda like smartphones, all it took is someone putting the pieces in the right order for the right purpose to make something more useful than the sum of its parts.

    • @michaeljburt
      @michaeljburt ปีที่แล้ว +11

      @@Alex-ye8qp To me this is the most bizarre part. How can a network even be trained to do that? So utterly bizarre to me. You're telling me "green cow grazing on mars" has a deterministic noise profile??

  • @patu8010
    @patu8010 ปีที่แล้ว +199

    I would love to hear more about the process. Like how does it recognize that the image now looks like a frog on stilts? Seems to me like that's where the real complexity is.

    • @skirtsonsale
      @skirtsonsale ปีที่แล้ว +36

      Same, I understood the noise subtraction bit, but I can't quite understand how the subtraction can lead to a picture of a frog, was the IA trained with "words vs images"? So it can relate what a frog would look like.
      Also, what the initial input picture (12:30) looks like? Is it just random generated noise?

    • @tristanstevens6162
      @tristanstevens6162 ปีที่แล้ว +53

      ​@@skirtsonsale yes a labeled dataset (image-text pairs, LAION dataset) was used to train the network. That is why it is called guided diffusion. The text guides the diffusion process not to a random image, but conditioned on the text (again the pairs were used for training).
      During training, it sample randoms noise from a random t according to the noise schedule (such that during training it is learned for all t). The input image on 12:30 is such image corrupted using noise from a random t. So somewhere between noise and an image.

    • @Darth_Bateman
      @Darth_Bateman ปีที่แล้ว

      @@skirtsonsale The concept of transformation in graphics will help you understand this.

    • @jonatansexdoer96
      @jonatansexdoer96 ปีที่แล้ว +17

      @@tristanstevens6162 But why doesn't that just produce some incohesive amalgamation of the training data? How does it know to specifically put the bunny ears on the frog's head? Is that where the magic of having a large amount training data comes in, in that it better understands the correlation between the label and the image?

    • @IceMetalPunk
      @IceMetalPunk ปีที่แล้ว +26

      @@jonatansexdoer96 Yep. CLIP is the language model used in these, and it's seen enough examples of things labeled "bunny" that look different from each other to abstract the idea of where bunny ears are located in any given underlying image.

  • @MikeSieko17
    @MikeSieko17 หลายเดือนก่อน

    so it's predicting the noise, and the randomness of the image comes from the random noise that the model gets? so if the model by chance got the same noise it would produce the same image? or something very close? or does each step also introduce randomness?

  • @troyrenaux9062
    @troyrenaux9062 ปีที่แล้ว

    When the network is producing an image with embedded text and producing two versions, one with and one without, is it doing both versions for every iteration down to t:0 or just using the difference calculated at t:T?

  • @user__214
    @user__214 ปีที่แล้ว +15

    My favorite part is where he explains AI while drawing on printer paper from 1989 XD

  • @mohammadmousavi1
    @mohammadmousavi1 ปีที่แล้ว

    Eagerly waiting for the deep dive video👍👍👍

  • @chyldstudios
    @chyldstudios ปีที่แล้ว +6

    Please do a walk-through of the Stable Diffusion code.

  • @1Chitus
    @1Chitus ปีที่แล้ว +1

    Where can I find his other videos where he is delving into the code etc? He explains very good.

  • @jytou
    @jytou 18 วันที่ผ่านมา

    Excellent explanation, thanks!

  • @monkeygame7
    @monkeygame7 ปีที่แล้ว

    So like, does the training data also include words associated with the images? e.g. when training with that picture of the rabbit, does it include "rabbit" or maybe some other modifiers?

  • @gorgolyt
    @gorgolyt ปีที่แล้ว +3

    When you say you remove the noise and then add most of it back, do you mean you add back the predicted noise, or you add back newly generated noise? And both approaches seem plausible, so what is the reasoning?

  • @smivan.
    @smivan. ปีที่แล้ว

    Thank you for covering this topic!

  • @zahidullah1707
    @zahidullah1707 5 หลายเดือนก่อน

    concise explanation Sir but why do we add noise to the original image?

  • @thejll
    @thejll 2 หลายเดือนก่อน

    Great video. Is there a name for the iterative denoising procedure ?

  • @TheDRAGONFLITE
    @TheDRAGONFLITE ปีที่แล้ว

    Please do that next video!!! You guys are great.

  • @AlanRuic
    @AlanRuic ปีที่แล้ว +8

    Could you add foreign language subtitles to these videos? Right now I'd need German subtitles. I've always loved your explainers but AI image generators is the first time where I'd need to show a video of yours to someone who's not fluent in English. TH-cam offers options to crowdsource subtitles BTW. Thanks a lot, keep up the good work! ☺

    • @phat80
      @phat80 ปีที่แล้ว

      TH-cam has a function of automatic subtitles.

  • @seralegre
    @seralegre ปีที่แล้ว

    love the high quality videos this channel always deliver

  • @sevret313
    @sevret313 ปีที่แล้ว +2

    What parameters did you use to generate those frog image? Tried it on my own local instance of StableDiffision and only got frogs not on stilts.

    • @IceMetalPunk
      @IceMetalPunk ปีที่แล้ว +2

      Try turning up the CFG scale. In general, higher values for that make the result match your prompt more closely, though it also reduces variation. If you're using the WebUI version, it also supports term emphasis, so you can try putting the word "stilts" in ((double parentheses)) to indicate that they should have higher weight in the result.

  • @Kram1032
    @Kram1032 ปีที่แล้ว +4

    Stable Diffusion is actually runable (in inference mode, i.e. for generation - this is different from training) on a regularish computer. The main factor is time but if you have a reasonably modern graphics card, you probably can run stable diffusion in principle.
    It just might take minutes rather than seconds for a single image.
    Somebody ran a variant of it on a not even that new iphone. It did take like half an hour iirc so it's not a thing most people would *want* to do, but one of the big selling poitns of stable diffusion is, that it's never the less *possible.*
    Stuff like Dall-E 2 or Imagen actually really does need a beefy computer with lots of specialized hardware (in particular, above-consumer-hardware VRAM) to get things done.
    Some of the oldest methods, though, can also work on a regular computer. I'm directly optimizing a version of CLIP towards some image for instance. It's not nearly as good as stable diffusion, but it's basically how all this madness of arbitrary images from text began

    • @asdf30111
      @asdf30111 ปีที่แล้ว +1

      The biggest limit is vram not gpu power as much, on a rtx 3000 series gpu 2-10s for 512^2 20 steps and 15-60 secs for 1024^2 image. Slow down starts due to vram not being able to hold everything. So you can get silly things like the 3060 being better then a 3080ti for some uses of SD.

    • @Kram1032
      @Kram1032 ปีที่แล้ว +1

      @@asdf30111 yeah VRAM is always the main issue with AI. Gotta store huge matrices in memory. I wonder if, going forward, hardware providers will bump up VRAM on their high end consumer cards due to increasing consumer demand...
      Or perhaps decently sized tensor cores will become more commonplace

  • @tristanwegner
    @tristanwegner ปีที่แล้ว

    The suggested follow up with details of the program would be great!

  • @lifeisbeautifu1
    @lifeisbeautifu1 6 หลายเดือนก่อน

    Amazing explanation. Thank you!

  • @TheAMadMan
    @TheAMadMan ปีที่แล้ว

    "That next video" sounds like exactly the video I want for these networks!

  • @amr-keleg
    @amr-keleg ปีที่แล้ว +2

    I really liked the explanation. Clear and easy to grasp. Thanks!

  • @luis1118
    @luis1118 ปีที่แล้ว +9

    Great video!! Would be interesting to see the differences between dall-e ans stable Diffusion. And how the last one requires less training and compute power

  • @adegboyegasamuel5190
    @adegboyegasamuel5190 7 หลายเดือนก่อน

    Computer Phile, this is very Good and intuitive 😊

  • @user-md6dj8ky6l
    @user-md6dj8ky6l 11 หลายเดือนก่อน

    А есть какое нибудь расширение, чтобы помогало писать prompt. Суфлер или на подобии т9, чтобы предлагало не кольцо вариантов написания слов/словосочетаний?

  • @jonathangrey9365
    @jonathangrey9365 5 หลายเดือนก่อน

    You did a great job explaining how the process works and provided visual examples. Nice work with this video.

  • @weekendstuff
    @weekendstuff ปีที่แล้ว

    Interesting to see (literally). Thanks for sharing.

  • @charlottehancocks5463
    @charlottehancocks5463 5 หลายเดือนก่อน

    I wish I could understand this. You must be a genius!

  • @naxey
    @naxey ปีที่แล้ว +1

    15:15 Why would we amplify the difference and not the parts that stay the same? Might be in the wrong here but in my understanding we are trying to use the parts that both the text guided and unguided noise produced to get the best output since the parts they both produced will be the best fit. Why then use the difference

  • @michipeka9973
    @michipeka9973 ปีที่แล้ว

    Fantastic explanation!

  • @user-qb5ky4qw1i
    @user-qb5ky4qw1i 10 หลายเดือนก่อน

    And by "giving it the noise estimate" are we referring to doing that via the text prompts?

  • @DHAiRYA2801
    @DHAiRYA2801 ปีที่แล้ว +2

    Thank you for this, would definitely love to see you going over the code and playing with this technique in the next video!

  • @devmuhnnad6425
    @devmuhnnad6425 11 หลายเดือนก่อน +1

    So Stable diffusion is a for loop at the end of the day, impressive 😌

  • @EU_DHD
    @EU_DHD ปีที่แล้ว +2

    New camera/processing? looks good!

  • @mckidney1
    @mckidney1 ปีที่แล้ว

    Follow-up video is going to be great!

  • @bosonh7723
    @bosonh7723 11 หลายเดือนก่อน +1

    Would it be possible to create videos from texts through diffusion? I mean I will write a story and it will generate a video of that story, what do you think we need to do for creating text to videos?

    • @SnoopyDoofie
      @SnoopyDoofie 10 หลายเดือนก่อน

      Google has been working on this and you can see an example in the 60 Minutes video they did on AI.

  • @lianemarie9280
    @lianemarie9280 18 วันที่ผ่านมา

    Always nice to listen to a lefty :)

  • @IbnIbrahem
    @IbnIbrahem ปีที่แล้ว

    How do you train the network with the text embeddings?

  • @anosognosiaanosmia1289
    @anosognosiaanosmia1289 ปีที่แล้ว +2

    Love this channel, it even helped me with IT certifications. Diffie Hellman for the win!!!

  • @dennisrkb
    @dennisrkb ปีที่แล้ว

    How do you train the embedding part?

  • @user-fp5cs9yf2z
    @user-fp5cs9yf2z 10 หลายเดือนก่อน

    Is there a way to estimate what percentage of the generated image is "of" the original image? Is it backing its way towards a single image, or a multitude of images?

    • @SnoopyDoofie
      @SnoopyDoofie 10 หลายเดือนก่อน

      It depends. If you provide the same prompt text and the same seed value, you'll always get the same image. If you change the seed value, you'll always get a different image. The seed has to do with how the initial image is initialized with random noise. The seed will set the pattern of the noise. In the video, he didn't mention anything about the initial image used when creating a new image. He mostly talked about the training process.

  • @deletedaxiom6057
    @deletedaxiom6057 ปีที่แล้ว

    On a side note: Is there some supply back room full of that old continuous stationary somewhere on campus? Good job recycling it, since there is no other use for it anymore.

  • @timetravellingblockhead2122
    @timetravellingblockhead2122 ปีที่แล้ว +2

    Thank you, now I can actually understand what the methods behind these generators are all about

  • @silviolasmar802
    @silviolasmar802 6 หลายเดือนก่อน

    Hello, Dall-E and MidJourney can be considered a DCGANs?

  • @achmedabadoba5478
    @achmedabadoba5478 ปีที่แล้ว

    Do you have a new/better camera? looks decent

  • @androidgameplays4every13
    @androidgameplays4every13 9 หลายเดือนก่อน

    Is this the same principle used by RVC AI voice cloning/generator?

  • @constantinosschinas4503
    @constantinosschinas4503 ปีที่แล้ว

    Love that he draws on old dot matrix paper.

  • @skyscraperfan
    @skyscraperfan ปีที่แล้ว +3

    How do I know what noise was added at which step, if both noises have the same frequency? If I start from the back with totally random noise and then subtract random noise, how can the result have less noise? If the noise is really random, it is no difference if you add or subtract it, because random noise is not any more or less random than its negative.

    • @MrRumcajs1000
      @MrRumcajs1000 ปีที่แล้ว +1

      if you randomly generate a black dot on a screen and then detect it, you can remove it through some kind of filter. It's true that it wouldn't technically be a subtraction, but I think that the idea stands.

  • @gaufqwi
    @gaufqwi ปีที่แล้ว +3

    As others have said, I'd love to see another video that elaborates on how the text prompt is factored into the model.

  • @Shabazza84
    @Shabazza84 6 หลายเดือนก่อน +9

    This is how DALL-E works in a nutshell:
    "Read user prompt. Decide it's against their arbitrary moral codex. Emit error."
    Excellent vid btw. Explained something complex in a very easy way.

  • @ilkero1067
    @ilkero1067 ปีที่แล้ว

    Can you elaborate on the embedding part?

  • @jasonschannel
    @jasonschannel ปีที่แล้ว +1

    i am a simple man. i see Mike Pound, i click

  • @soulaymanal-abdallah6410
    @soulaymanal-abdallah6410 ปีที่แล้ว

    amazing explanation!

  • @housellama
    @housellama ปีที่แล้ว +4

    "I saw the angel in the marble and carved until I set him free. ” - Michelangelo

  • @mullhamqattan8551
    @mullhamqattan8551 ปีที่แล้ว

    can i get a link of the paper that contains these details???

  • @p0wnyman
    @p0wnyman ปีที่แล้ว +19

    Mate, that '8 pounds' had me, I love your work, very informative and entertaining, cheers.

  • @naxey
    @naxey ปีที่แล้ว +1

    Just starts trowing loose change at him at the end😂😂
    Jokes aside, loved the video, I finally somewhat grasp how this black magic works

  • @WALLACE9009
    @WALLACE9009 11 หลายเดือนก่อน

    It is amazing that something like that works

  • @amesino.i
    @amesino.i ปีที่แล้ว

    Great explanation. Thanks.

  • @jevaispartout1271
    @jevaispartout1271 ปีที่แล้ว

    How many stacks of perforated papers in your stock?

  • @jimsmind3894
    @jimsmind3894 ปีที่แล้ว

    I'm curious as to whether quantum computing will speed up this sort of process?
    It sounds like the sort of weighted vector probability math that would be optimal on a q-bit based computer?

    • @IceMetalPunk
      @IceMetalPunk ปีที่แล้ว +1

      Basically all neural network training relies on lots of linear algebra (mostly matrix multiplication), and from what I've read in a *very* brief lookup just now, the HHL quantum algorithm is known to perform linear algebra much faster than classical computers. And we've got a working demo of it already as of 2018 (though for the small number of qubits that our current quantum computers are limited to). So in theory, yeah, once we have scalable enough quantum computers with enough qubits, most modern AI could be made far faster with HHL.

  • @pwrtrace
    @pwrtrace ปีที่แล้ว

    is there any way to write a text in the image generated?

  • @GregMoress
    @GregMoress ปีที่แล้ว +2

    Two things amaze me... First, the AI-aspect which I will need more time to study (it's new to me).
    Second: Mate... they still make folded printer paper like that? It's been decades since I last saw it. You're near a mainframe ammirite?

    • @salomeshunamon4737
      @salomeshunamon4737 ปีที่แล้ว +1

      I went looking on amazon for it after watching. LOL

  • @icu9313
    @icu9313 ปีที่แล้ว

    6:30
    Isn't this concept the same as solving P=NP ?
    Find the question only using the answer?

  • @MarkCastle
    @MarkCastle 11 หลายเดือนก่อน

    Love that he’s drawing on dotmatrix printer paper !!

  • @sambeard4428
    @sambeard4428 ปีที่แล้ว +18

    I feel the GPT embedding is skipped over quite easily. With my limited knowledge on how these text Transformers work there is a vector representation of the description that in stead of representing the words represent the 'meaning' of the text. However how does this transform to an image that does exactly that? You would still need a lot of training data that confirms whether an image matches a description, right? Did they use alt descriptions or something similar for this, eg (publicly) available image descriptions? I guess my point is that i do not see how this 'knowledge' of what's in the text is transferred to the 'knowledge' of what's in the image, apart from there being a mechanism of steering it towards the 'knowledge' but where does this knowledge come from?

    • @gooblepls3985
      @gooblepls3985 ปีที่แล้ว +7

      The network is basically implicitly imbued with this understanding through the training process:
      For the training process of these conditional models, you take a labeled dataset {(image, label)}, and the label is a natural-language description text *describing that image* . For a particular data sample, the network will then see a noisy version of this image, along with the embedding of the description of the original clean image. The training loss then rewards the network for reproducing this image given these two bits of information, and punishes it for reproducing *any* other image. I hope it's clear that the training will therefore steer the network into a direction of 'understanding' these descriptions in some abstract way.
      That this understanding of the input text is *abstract* is actually what you hope allows the model to extrapolate, so it can generate new images conditioned on new texts not seen during training. That this extrapolation works ridiculously well is one of the amazing things about these models.

  • @FidozIsGod
    @FidozIsGod ปีที่แล้ว

    What a wonderful video. Thank you for making this!

  • @chriswatts3697
    @chriswatts3697 8 หลายเดือนก่อน

    Why Noise? Because it is something like a pixelated disintegration of the image that can be stored mathematically? So it's easier to compare its structure with other images?

  • @manuelnovella39
    @manuelnovella39 ปีที่แล้ว +5

    Videos about AI are so cool. Please, more!

  • @runakovacs4759
    @runakovacs4759 ปีที่แล้ว

    Yay new Mike Pound video

  • @DampeS8N
    @DampeS8N ปีที่แล้ว +6

    I'm running Stable Diffusion locally on a 3080 Ti, works fine.

    • @MattFowlerBTR
      @MattFowlerBTR ปีที่แล้ว +2

      Gaming laptop with a 3060 (6GB) here working great for Stable Diffusion, I'm using the Automatic1111 web UI distro bundle which made setup incredibly easy. I'm still learning how to use it to get "what I mean" results, but it is quite amazing.

    • @edgex4239
      @edgex4239 ปีที่แล้ว

      I've ran it on a 2060 fine, though I think its the lowest supported card

    • @IceMetalPunk
      @IceMetalPunk ปีที่แล้ว +1

      I'm using a 3060 Ti. But inference is *always* more efficient than training, so my hardware can't handle Dreambooth, and certainly would never come close to handling a full initial training.