Loading video...

Video Failed to Load

There was a problem loading this video. This could be due to a temporary network issue or the video might be unavailable.

LTX-2.5 is native in ComfyUI, day zero with the weights Model highlights: → Diffusion Fidelity Rendering - compute allocated by scene complexity, not spread evenly → Diffusion Video Decoder - replaces VAE decoding, sharper faces and legible text → Native multi-shot - one run, multiple connected shots, consistent across... show more

ComfyUI

54,693 subscribers

17,791 views • 1 day ago •via X (Twitter)

Arts Science & Technology Education

Anya Rossi• Live Now

Private livecam show

0 Comments

No comments available

Comments from the original post will appear here

Related Videos

FLUX 3 Video is live in ComfyUI via Partner Nodes Black Forest Labs FLUX 3 Model highlights: → 20s in a single generation → native audio, not dubbed on after → text-to-video, image-to-video, keyframes, continuation → multi-scene and multi-angle in one prompt → multilingual dialogue with lipsync that holds → typography that sits in the scene → chain clips agentically for longer stories One model across video, audio, image, and action. Click the link below to download the workflows and learn more with blog

FLUX 3 Video is live in ComfyUI via Partner Nodes Black Forest Labs FLUX 3 Model highlights: → 20s in a single generation → native audio, not dubbed on after → text-to-video, image-to-video, keyframes, continuation → multi-scene and multi-angle in one prompt → multilingual dialogue with lipsync that holds → typography that sits in the scene → chain clips agentically for longer stories One model across video, audio, image, and action. Click the link below to download the workflows and learn more with blog

ComfyUI

24,656 views • 7 days ago

MiniMax H3 is native in ComfyUI, day zero with the weights. Model highlights: → Text-to-video, prompt only → Image-to-video → First-and-last-frame - control the opening frame, the closing frame, or both → Reference-to-video - carry a subject, a motion, or a voice through the clip → In-place editing - modify a shot you already have instead of regenerating it The capability MiniMax (official) leads with is multimodal context understanding, and it's what collapses those five tasks into one model. H3 takes images, audio, and video together and resolves them against a prompt that explains how they relate. Download the workflows & learn more in our blog 👇

MiniMax H3 is native in ComfyUI, day zero with the weights. Model highlights: → Text-to-video, prompt only → Image-to-video → First-and-last-frame - control the opening frame, the closing frame, or both → Reference-to-video - carry a subject, a motion, or a voice through the clip → In-place editing - modify a shot you already have instead of regenerating it The capability MiniMax (official) leads with is multimodal context understanding, and it's what collapses those five tasks into one model. H3 takes images, audio, and video together and resolves them against a prompt that explains how they relate. Download the workflows & learn more in our blog 👇

ComfyUI

226,164 views • 10 days ago

Seedream 5.0 Pro is now in ComfyUI via Partner Nodes. ByteDance's latest image model brings: → Character & product consistency across edits → Precision region editing → Infographics, flowcharts & structured layouts → In-image text across 14 languages One model. Text-to-image, editing, and multi-image inputs. Click the link below for the workflow 👇

Seedream 5.0 Pro is now in ComfyUI via Partner Nodes. ByteDance's latest image model brings: → Character & product consistency across edits → Precision region editing → Infographics, flowcharts & structured layouts → In-image text across 14 languages One model. Text-to-image, editing, and multi-image inputs. Click the link below for the workflow 👇

ComfyUI

14,756 views • 1 month ago

Upscale-A-Video: Temporal-Consistent Diffusion Model for Real-World Video Super-Resolution paper page: Text-based diffusion models have exhibited remarkable success in generation and editing, showing great promise for enhancing visual content with their generative prior. However, applying these models to video super-resolution remains challenging due to the high demands for output fidelity and temporal consistency, which is complicated by the inherent randomness in diffusion models. Our study introduces Upscale-A-Video, a text-guided latent diffusion framework for video upscaling. This framework ensures temporal coherence through two key mechanisms: locally, it integrates temporal layers into U-Net and VAE-Decoder, maintaining consistency within short sequences; globally, without training, a flow-guided recurrent latent propagation module is introduced to enhance overall video stability by propagating and fusing latent across the entire sequences. Thanks to the diffusion paradigm, our model also offers greater flexibility by allowing text prompts to guide texture creation and adjustable noise levels to balance restoration and generation, enabling a trade-off between fidelity and quality. Extensive experiments show that Upscale-A-Video surpasses existing methods in both synthetic and real-world benchmarks, as well as in AI-generated videos, showcasing impressive visual realism and temporal consistency.

Upscale-A-Video: Temporal-Consistent Diffusion Model for Real-World Video Super-Resolution paper page: Text-based diffusion models have exhibited remarkable success in generation and editing, showing great promise for enhancing visual content with their generative prior. However, applying these models to video super-resolution remains challenging due to the high demands for output fidelity and temporal consistency, which is complicated by the inherent randomness in diffusion models. Our study introduces Upscale-A-Video, a text-guided latent diffusion framework for video upscaling. This framework ensures temporal coherence through two key mechanisms: locally, it integrates temporal layers into U-Net and VAE-Decoder, maintaining consistency within short sequences; globally, without training, a flow-guided recurrent latent propagation module is introduced to enhance overall video stability by propagating and fusing latent across the entire sequences. Thanks to the diffusion paradigm, our model also offers greater flexibility by allowing text prompts to guide texture creation and adjustable noise levels to balance restoration and generation, enabling a trade-off between fidelity and quality. Extensive experiments show that Upscale-A-Video surpasses existing methods in both synthetic and real-world benchmarks, as well as in AI-generated videos, showcasing impressive visual realism and temporal consistency.

AK

32,849 views • 2 years ago

Get Started with LTX-2 in ComfyUI The open-source model that turns prompts into full cinematic video. • Motion, dialogue, SFX, and music — generated together • Precise control with Canny, Depth & Pose inputs • Built-in upscaling and prompt enhancement

Get Started with LTX-2 in ComfyUI The open-source model that turns prompts into full cinematic video. • Motion, dialogue, SFX, and music — generated together • Precise control with Canny, Depth & Pose inputs • Built-in upscaling and prompt enhancement

ComfyUI

20,696 views • 7 months ago

$Auto regressive LLMs are officially on notice. run Gemma 4 26B diffusion gguf with llama.cpp Google just dropped DiffusionGemma-26B, and it completely flips how we generate text. instead of predicting words one by one, it generates 256 tokens in parallel using bi-directional attention. its like stable diffusion, but for language. the model starts with random text "noise" and iteratively refines and self-corrects the entire block in real-time to fix formatting and reasoning errors on the fly. since it’s a Mixture of Experts (MoE) that only activates 3.8B parameters during inference, it fits perfectly on consumer hardware. You can run the Q4_K_M quant with an 18GB VRAM budget on a single RTX 3090 or RTX 4090 with exceptional throughput. Tested on Ubuntu 22 with CUDA 13.1 using the cutting edge experimental llama.cpp branch. Here is how to compile and run it with the live terminal denoising visualizer: # 1. Clone & check out the experimental PR (#24423) - 1) git clone && cd llama.cpp -git fetch origin 2) pull/24423/head:diffusiongemma && --git checkout diffusiongemma # 2. Build with CUDA support 1) cmake -B build -DGGML_CUDA=ON -DCMAKE_CUDA_ARCHITECTURES=native 2) cmake --build build -j $(nproc) --config Release --target llama-diffusion-cli # 3. Run with live visual denoising (llama.cpp flags) ./build/bin/llama-diffusion-cli \ -m /path/to/diffusiongemma-26B-A4B-it-Q4_K_M.gguf \ -ngl 99 -cnv -n 2048 --diffusion-visual Watch the video below to see the live --diffusion-visual canvas iteratively de noising the prompt output in real time. guide and unsloth's hugging face GGUF model links are in the comments below! Is auto regressive generation officially legacy tech? Let me know what you think.$

Auto regressive LLMs are officially on notice. run Gemma 4 26B diffusion gguf with llama.cpp Google just dropped DiffusionGemma-26B, and it completely flips how we generate text. instead of predicting words one by one, it generates 256 tokens in parallel using bi-directional attention. its like stable diffusion, but for language. the model starts with random text "noise" and iteratively refines and self-corrects the entire block in real-time to fix formatting and reasoning errors on the fly. since it’s a Mixture of Experts (MoE) that only activates 3.8B parameters during inference, it fits perfectly on consumer hardware. You can run the Q4_K_M quant with an 18GB VRAM budget on a single RTX 3090 or RTX 4090 with exceptional throughput. Tested on Ubuntu 22 with CUDA 13.1 using the cutting edge experimental llama.cpp branch. Here is how to compile and run it with the live terminal denoising visualizer: # 1. Clone & check out the experimental PR (#24423) - 1) git clone && cd llama.cpp -git fetch origin 2) pull/24423/head:diffusiongemma && --git checkout diffusiongemma # 2. Build with CUDA support 1) cmake -B build -DGGML_CUDA=ON -DCMAKE_CUDA_ARCHITECTURES=native 2) cmake --build build -j $(nproc) --config Release --target llama-diffusion-cli # 3. Run with live visual denoising (llama.cpp flags) ./build/bin/llama-diffusion-cli \ -m /path/to/diffusiongemma-26B-A4B-it-Q4_K_M.gguf \ -ngl 99 -cnv -n 2048 --diffusion-visual Watch the video below to see the live --diffusion-visual canvas iteratively de noising the prompt output in real time. guide and unsloth's hugging face GGUF model links are in the comments below! Is auto regressive generation officially legacy tech? Let me know what you think.

Alok

52,656 views • 2 months ago

We've officially released and open-sourced HunyuanImage 2.1, our latest text-to-image model. The new model delivers on our commitment to balancing performance and quality. With native 2K image generation, HunyuanImage 2.1 is an advanced open-source text-to-image model.🎨 ✨ New in 2.1: 🔹Advanced Semantics: Supports ultra-long and complex prompts of up to 1000 tokens, and precisely controls the generation of multiple subjects in a single image. 🔹Precise Chinese and English Text Rendering with seamless image–text integration: The model naturally integrates text into images, making it suitable for a wide range of applications such as product covers, illustrations, and poster design to meet the needs of various fields. 🔹Rich Styles and High Aesthetic: Capable of generating images in various styles—including photorealistic portraits, comics, and vinyl figures—it delivers outstanding visual appeal and artistic quality. 🔹High-Quality Generation: Efficiently produces ultra-high-definition (2K) images in the same time other models take to generate a 1K image. HunyuanImage 2.1 uses two text encoders: a multimodal large language model (MLLM) to improve the model's image and text alignment capabilities, and a multi-language character-aware encoder to improve text rendering capabilities. The model is a single- and double-stream diffusion transformer with 17B parameters. We've also open-sourced the weights of the the accelerated version with meanflow which reduces inference steps from 100 to just 8, and PromptEnhancer, the first industrial-grade rewriting model that enhances your prompts for more nuanced and expressive image generation. Now, creators turn complex ideas—like posters with slogans or multi-panel comics—into visuals faster than ever. We’re just getting started. Stay tuned for our native multimodal image generation model coming soon. 🌐Website: 🔗Github: 🤗Hugging Face: ✨Hugging Face Demo:

We've officially released and open-sourced HunyuanImage 2.1, our latest text-to-image model. The new model delivers on our commitment to balancing performance and quality. With native 2K image generation, HunyuanImage 2.1 is an advanced open-source text-to-image model.🎨 ✨ New in 2.1: 🔹Advanced Semantics: Supports ultra-long and complex prompts of up to 1000 tokens, and precisely controls the generation of multiple subjects in a single image. 🔹Precise Chinese and English Text Rendering with seamless image–text integration: The model naturally integrates text into images, making it suitable for a wide range of applications such as product covers, illustrations, and poster design to meet the needs of various fields. 🔹Rich Styles and High Aesthetic: Capable of generating images in various styles—including photorealistic portraits, comics, and vinyl figures—it delivers outstanding visual appeal and artistic quality. 🔹High-Quality Generation: Efficiently produces ultra-high-definition (2K) images in the same time other models take to generate a 1K image. HunyuanImage 2.1 uses two text encoders: a multimodal large language model (MLLM) to improve the model's image and text alignment capabilities, and a multi-language character-aware encoder to improve text rendering capabilities. The model is a single- and double-stream diffusion transformer with 17B parameters. We've also open-sourced the weights of the the accelerated version with meanflow which reduces inference steps from 100 to just 8, and PromptEnhancer, the first industrial-grade rewriting model that enhances your prompts for more nuanced and expressive image generation. Now, creators turn complex ideas—like posters with slogans or multi-panel comics—into visuals faster than ever. We’re just getting started. Stay tuned for our native multimodal image generation model coming soon. 🌐Website: 🔗Github: 🤗Hugging Face: ✨Hugging Face Demo:

Tencent Hy

89,257 views • 11 months ago

TurboEdit Instant text-based image editing discuss: We address the challenges of precise image inversion and disentangled image editing in the context of few-step diffusion models. We introduce an encoder based iterative inversion technique. The inversion network is conditioned on the input image and the reconstructed image from the previous step, allowing for correction of the next reconstruction towards the input image. We demonstrate that disentangled controls can be easily achieved in the few-step diffusion model by conditioning on an (automatically generated) detailed text prompt. To manipulate the inverted image, we freeze the noise maps and modify one attribute in the text prompt (either manually or via instruction based editing driven by an LLM), resulting in the generation of a new image similar to the input image with only one attribute changed. It can further control the editing strength and accept instructive text prompt. Our approach facilitates realistic text-guided image edits in real-time, requiring only 8 number of functional evaluations (NFEs) in inversion (one-time cost) and 4 NFEs per edit. Our method is not only fast, but also significantly outperforms state-of-the-art multi-step diffusion editing techniques.

TurboEdit Instant text-based image editing discuss: We address the challenges of precise image inversion and disentangled image editing in the context of few-step diffusion models. We introduce an encoder based iterative inversion technique. The inversion network is conditioned on the input image and the reconstructed image from the previous step, allowing for correction of the next reconstruction towards the input image. We demonstrate that disentangled controls can be easily achieved in the few-step diffusion model by conditioning on an (automatically generated) detailed text prompt. To manipulate the inverted image, we freeze the noise maps and modify one attribute in the text prompt (either manually or via instruction based editing driven by an LLM), resulting in the generation of a new image similar to the input image with only one attribute changed. It can further control the editing strength and accept instructive text prompt. Our approach facilitates realistic text-guided image edits in real-time, requiring only 8 number of functional evaluations (NFEs) in inversion (one-time cost) and 4 NFEs per edit. Our method is not only fast, but also significantly outperforms state-of-the-art multi-step diffusion editing techniques.

AK

16,062 views • 2 years ago

LTX-2 is natively supported in ComfyUI on Day 0 🎬🔊 The next chapter in controllability for open-source video generation. - Open-source audio-video foundation model - Generates motion, dialogue, SFX, and music together - Canny, Depth & Pose video-to-video control - Keyframe-driven generation - Native upscaling and prompt enhancement Everything in a single pass.

LTX-2 is natively supported in ComfyUI on Day 0 🎬🔊 The next chapter in controllability for open-source video generation. - Open-source audio-video foundation model - Generates motion, dialogue, SFX, and music together - Canny, Depth & Pose video-to-video control - Keyframe-driven generation - Native upscaling and prompt enhancement Everything in a single pass.

ComfyUI

108,333 views • 7 months ago

Stable Diffusion Moment of Audio?? Ace-Step Audio Model just dropped and is now natively supported in ComfyUI. ACE-Step is an open-source music generation model jointly developed by ACE Studio and StepFun.

Stable Diffusion Moment of Audio?? Ace-Step Audio Model just dropped and is now natively supported in ComfyUI. ACE-Step is an open-source music generation model jointly developed by ACE Studio and StepFun.

ComfyUI

23,611 views • 1 year ago

ProlificDreamer: High-Fidelity and Diverse Text-to-3D Generation with Variational Score Distillation propose to model the 3D parameter as a random variable instead of a constant as in SDS and present variational score distillation (VSD), a principled particle-based variational framework to explain and address the aforementioned issues in text-to-3D generation. We show that SDS is a special case of VSD and leads to poor samples with both small and large CFG weights. In comparison, VSD works well with various CFG weights as ancestral sampling from diffusion models and simultaneously improves the diversity and sample quality with a common CFG weight (i.e., 7.5). We further present various improvements in the design space for text-to-3D such as distillation time schedule and density initialization, which are orthogonal to the distillation algorithm yet not well explored. Our overall approach, dubbed ProlificDreamer, can generate high rendering resolution (i.e., 512times512) and high-fidelity NeRF with rich structure and complex effects (e.g., smoke and drops). Further, initialized from NeRF, meshes fine-tuned by VSD are meticulously detailed and photo-realistic paper page:

ProlificDreamer: High-Fidelity and Diverse Text-to-3D Generation with Variational Score Distillation propose to model the 3D parameter as a random variable instead of a constant as in SDS and present variational score distillation (VSD), a principled particle-based variational framework to explain and address the aforementioned issues in text-to-3D generation. We show that SDS is a special case of VSD and leads to poor samples with both small and large CFG weights. In comparison, VSD works well with various CFG weights as ancestral sampling from diffusion models and simultaneously improves the diversity and sample quality with a common CFG weight (i.e., 7.5). We further present various improvements in the design space for text-to-3D such as distillation time schedule and density initialization, which are orthogonal to the distillation algorithm yet not well explored. Our overall approach, dubbed ProlificDreamer, can generate high rendering resolution (i.e., 512times512) and high-fidelity NeRF with rich structure and complex effects (e.g., smoke and drops). Further, initialized from NeRF, meshes fine-tuned by VSD are meticulously detailed and photo-realistic paper page:

AK

46,151 views • 3 years ago

Face Replacement and De-aging Demo: VFX artist Doug Hogan built a fully automated face swap workflow inside ComfyUI. Florence 2 → SAM2 → WAN Video. One pipeline. Temporally consistent across every frame. Not a pixel swap. The workflow detects the head, segments it, drives motion from the original performance, and uses a reference image + auto prompt to transfer the identity while preserving lighting and timing. Face replacement. De-aging. All inside ComfyUI. Click the link below for the workflow 👇

Face Replacement and De-aging Demo: VFX artist Doug Hogan built a fully automated face swap workflow inside ComfyUI. Florence 2 → SAM2 → WAN Video. One pipeline. Temporally consistent across every frame. Not a pixel swap. The workflow detects the head, segments it, drives motion from the original performance, and uses a reference image + auto prompt to transfer the identity while preserving lighting and timing. Face replacement. De-aging. All inside ComfyUI. Click the link below for the workflow 👇

ComfyUI

13,948 views • 1 month ago

🚨 BREAKING: A research lab just released a 15B model that generates multilingual talking human videos with synced audio, beats every competitor in human evaluation, and runs in 38 seconds on one GPU. It's called daVinci-MagiHuman. The key insight is that every other model in this category stacks cross-attention, multi-stream pipelines, and separate conditioning branches to handle video and audio together. This one throws all of that out and uses a single unified self-attention stream across all modalities. Super-resolution happens in latent space rather than pixel space so there's no extra VAE decode-encode round trip. The turbo VAE decoder cuts decoding overhead even further. The distilled version runs in 8 steps with no CFG at all. Visual quality, text alignment, and word error rate all beat Ovi 1.1 and LTX 2.3 on the benchmark table. 100% Opensource. Apache 2.0. Repo and research paper links are in the comments.

🚨 BREAKING: A research lab just released a 15B model that generates multilingual talking human videos with synced audio, beats every competitor in human evaluation, and runs in 38 seconds on one GPU. It's called daVinci-MagiHuman. The key insight is that every other model in this category stacks cross-attention, multi-stream pipelines, and separate conditioning branches to handle video and audio together. This one throws all of that out and uses a single unified self-attention stream across all modalities. Super-resolution happens in latent space rather than pixel space so there's no extra VAE decode-encode round trip. The turbo VAE decoder cuts decoding overhead even further. The distilled version runs in 8 steps with no CFG at all. Visual quality, text alignment, and word error rate all beat Ovi 1.1 and LTX 2.3 on the benchmark table. 100% Opensource. Apache 2.0. Repo and research paper links are in the comments.

Ihtesham Ali

45,579 views • 4 months ago

Frame Adjustments Demo: Most people think AI filmmaking means finding the one perfect model. It doesn't. Doug Hogan breaks down why the best workflows are mix-and-match. Different models for different shots, with LLMs writing your prompts and adapting them per-model automatically.

Frame Adjustments Demo: Most people think AI filmmaking means finding the one perfect model. It doesn't. Doug Hogan breaks down why the best workflows are mix-and-match. Different models for different shots, with LLMs writing your prompts and adapting them per-model automatically.

ComfyUI

13,299 views • 2 months ago

Seedance Playblast 2 Render Demo: VFX artist Doug Hogan built a workflow in ComfyUI that takes a raw Playblast and turns it into a finished shot using Seedance 2.0 - no traditional rendering pipeline. → Playblast = motion ground truth → Nanobanana Pro multi-angle grids = asset identity → Gemini node = auto-generated structured prompt → Seedance 2.0 ties it all together Raw CG animation → finished shot in minutes instead of hours. Click the link below for the workflow👇

Seedance Playblast 2 Render Demo: VFX artist Doug Hogan built a workflow in ComfyUI that takes a raw Playblast and turns it into a finished shot using Seedance 2.0 - no traditional rendering pipeline. → Playblast = motion ground truth → Nanobanana Pro multi-angle grids = asset identity → Gemini node = auto-generated structured prompt → Seedance 2.0 ties it all together Raw CG animation → finished shot in minutes instead of hours. Click the link below for the workflow👇

ComfyUI

13,957 views • 2 months ago

ComfyUI, even on vacation. 🌴 While on vacation, enigmatic_e left the laptop behind. A phone and an AI agent connected to ComfyUI MCP were all it took to run ComfyUI workflows through natural conversation while vacation continued. To try this workflow and learn more about ComfyUI MCP, check out the links below 👇

ComfyUI, even on vacation. 🌴 While on vacation, enigmatic_e left the laptop behind. A phone and an AI agent connected to ComfyUI MCP were all it took to run ComfyUI workflows through natural conversation while vacation continued. To try this workflow and learn more about ComfyUI MCP, check out the links below 👇

ComfyUI

123,383 views • 28 days ago

Comfy Desktop is here - one app for every Comfy. We rebuilt the desktop experience from the ground up. Your existing workflows, custom nodes, models, and settings carry over untouched. Same name, completely new app. What's new: → Multiple ComfyUI instances — different custom nodes, different versions, all managed in one place → Automatic snapshots before every update and custom node change, with one-click rollback if something breaks → Day-0 ComfyUI releases — the moment a new version or nightly drops, you can update immediately If you're on our older desktop app, you'll see an update prompt automatically.

Comfy Desktop is here - one app for every Comfy. We rebuilt the desktop experience from the ground up. Your existing workflows, custom nodes, models, and settings carry over untouched. Same name, completely new app. What's new: → Multiple ComfyUI instances — different custom nodes, different versions, all managed in one place → Automatic snapshots before every update and custom node change, with one-click rollback if something breaks → Day-0 ComfyUI releases — the moment a new version or nightly drops, you can update immediately If you're on our older desktop app, you'll see an update prompt automatically.

ComfyUI

80,769 views • 2 months ago

Vidu Q3 is now live in ComfyUI! The latest Vidu video generation model delivers stable, synchronized audio-visual performance to your workflow. - Synchronized audio & dialogue - Intelligent cinematography with auto/manual control - 1-16 second flexible duration - Multi-language support Create immersive videos with sound in one workflow. 👇 Try it now

Vidu Q3 is now live in ComfyUI! The latest Vidu video generation model delivers stable, synchronized audio-visual performance to your workflow. - Synchronized audio & dialogue - Intelligent cinematography with auto/manual control - 1-16 second flexible duration - Multi-language support Create immersive videos with sound in one workflow. 👇 Try it now

ComfyUI

14,147 views • 6 months ago

The Houdini-ComfyUI Bridge plugin is open-sourced 🎉 This community project by Rafael Drelich and Anatolii Iudanov helps bring ComfyUI’s generative AI abilities directly into Houdini’s node graph: - Load comfyUI nodes directly into COPs - Import/export from ComfyUI with ease: images, meshes, audio, etc. - TOPs submitter allows users to develop custom pipelines that combine CG with diffusion and generative AI

The Houdini-ComfyUI Bridge plugin is open-sourced 🎉 This community project by Rafael Drelich and Anatolii Iudanov helps bring ComfyUI’s generative AI abilities directly into Houdini’s node graph: - Load comfyUI nodes directly into COPs - Import/export from ComfyUI with ease: images, meshes, audio, etc. - TOPs submitter allows users to develop custom pipelines that combine CG with diffusion and generative AI

ComfyUI

21,423 views • 6 months ago

SORA by Hand ✍️ OpenAI’s #SORA took over the Internet when it was announced earlier this year. The technology behind Sora is the Diffusion Transformer (DiT) developed by William Peebles and Shining Xie. How does DiT work? 𝗚𝗼𝗮𝗹: Generate a video conditioned by a text prompt and a series of diffusion steps [1] Given ↳ Video ↳ Prompt: "sora is sky" ↳ Diffusion step: t = 3 [2] Video → Patches ↳ Divide all pixels in all frames into 4 spacetime patches [3] Visual Encoder: Pixels 🟨 → Latent 🟩 ↳ Multiply the patches with weights and biases, followed by ReLU ↳ The result is a latent feature vector per patch ↳ The purpose is dimension reduction from 4 (2x2x1) to 2 (2x1). ↳ In the paper, the reduction is 196,608 (256x256x3)→ 4096 (32x32x4) [4] ⬛ Add Noise ↳ Sample a noise according to the diffusion time step t. Typically, the larger the t, the smaller the noise. ↳ Add the Sampled Noise to latent features to obtain Noised Latent. ↳ The goal is to purposely add noise to a video and ask the model to guess what that noise is. ↳ This is analogous to training a language model by purposely deleting a word in a sentence and ask the model to guess what the deleted word was. [5-7] 🟪 Conditioning by Adaptive Layer Norm [5] Encode Conditions ↳ Encode "sora is sky" into a text embedding vector [0,1,-1]. ↳ Encode t = 3 to as a binary vector [1,1]. ↳ Concatenate the two vectors in to a 5D column vector. [6] Estimate Scale/Shift ↳ Multiply the combined vector with weights and biases ↳ The goal is to estimate the scale [2,-1] and shift [-1,5]. ↳ Copy the result to (X) and (+) [7] Apply Scale/Sift ↳ Scale the noised latent by [2,-1] ↳ Shifted the scaled noised latent by [-1, 5] ↳ The result is "conditioned" noise latent. [8-10] Transformer [8] Self-Attention ↳ Feed the conditioned noised latent to Query-Key function to obtain a self-attention matrix ↳ Value is omitted for simplicity [9] Attention Pooling ↳ Multiply the conditioned noised latent with the self-attention matrix ↳ The result are attention weighted features [10] Pointwise Feed Forward Network ↳ Multiply the attention weighted features with weights and biases ↳ The result is the Predicted Noise 🏋️‍♂️ 𝗧𝗿𝗮𝗶𝗻 [11] ↳ Calculate MSE loss gradients by taking the different between the Predicted Noise and the Sampled Noise (ground truth). ↳ Use the loss gradients to kick off backpropagation to update all learnable parameters (red borders) ↳ Note the visual encoder and decoder's parameters are frozen (blue borders) 🎨 𝗚𝗲𝗻𝗲𝗿𝗮𝘁𝗲 (𝗦𝗮𝗺𝗽𝗹𝗲) [12] Denoise ↳ Subtract the predicted noise from the noised latent to obtain the noise-free latent [13] Visual Decoder: Latent 🟩 → Pixels 🟨 ↳ Multiply the patches with weights and biases, followed by ReLU [14] Patches → Video ↳ Rearrange patches into a sequence of video frames.

SORA by Hand ✍️ OpenAI’s #SORA took over the Internet when it was announced earlier this year. The technology behind Sora is the Diffusion Transformer (DiT) developed by William Peebles and Shining Xie. How does DiT work? 𝗚𝗼𝗮𝗹: Generate a video conditioned by a text prompt and a series of diffusion steps [1] Given ↳ Video ↳ Prompt: "sora is sky" ↳ Diffusion step: t = 3 [2] Video → Patches ↳ Divide all pixels in all frames into 4 spacetime patches [3] Visual Encoder: Pixels 🟨 → Latent 🟩 ↳ Multiply the patches with weights and biases, followed by ReLU ↳ The result is a latent feature vector per patch ↳ The purpose is dimension reduction from 4 (2x2x1) to 2 (2x1). ↳ In the paper, the reduction is 196,608 (256x256x3)→ 4096 (32x32x4) [4] ⬛ Add Noise ↳ Sample a noise according to the diffusion time step t. Typically, the larger the t, the smaller the noise. ↳ Add the Sampled Noise to latent features to obtain Noised Latent. ↳ The goal is to purposely add noise to a video and ask the model to guess what that noise is. ↳ This is analogous to training a language model by purposely deleting a word in a sentence and ask the model to guess what the deleted word was. [5-7] 🟪 Conditioning by Adaptive Layer Norm [5] Encode Conditions ↳ Encode "sora is sky" into a text embedding vector [0,1,-1]. ↳ Encode t = 3 to as a binary vector [1,1]. ↳ Concatenate the two vectors in to a 5D column vector. [6] Estimate Scale/Shift ↳ Multiply the combined vector with weights and biases ↳ The goal is to estimate the scale [2,-1] and shift [-1,5]. ↳ Copy the result to (X) and (+) [7] Apply Scale/Sift ↳ Scale the noised latent by [2,-1] ↳ Shifted the scaled noised latent by [-1, 5] ↳ The result is "conditioned" noise latent. [8-10] Transformer [8] Self-Attention ↳ Feed the conditioned noised latent to Query-Key function to obtain a self-attention matrix ↳ Value is omitted for simplicity [9] Attention Pooling ↳ Multiply the conditioned noised latent with the self-attention matrix ↳ The result are attention weighted features [10] Pointwise Feed Forward Network ↳ Multiply the attention weighted features with weights and biases ↳ The result is the Predicted Noise 🏋️‍♂️ 𝗧𝗿𝗮𝗶𝗻 [11] ↳ Calculate MSE loss gradients by taking the different between the Predicted Noise and the Sampled Noise (ground truth). ↳ Use the loss gradients to kick off backpropagation to update all learnable parameters (red borders) ↳ Note the visual encoder and decoder's parameters are frozen (blue borders) 🎨 𝗚𝗲𝗻𝗲𝗿𝗮𝘁𝗲 (𝗦𝗮𝗺𝗽𝗹𝗲) [12] Denoise ↳ Subtract the predicted noise from the noised latent to obtain the noise-free latent [13] Visual Decoder: Latent 🟩 → Pixels 🟨 ↳ Multiply the patches with weights and biases, followed by ReLU [14] Patches → Video ↳ Rearrange patches into a sequence of video frames.

Tom Yeh

238,303 views • 2 years ago