Video wird geladen...

Video konnte nicht geladen werden

Beim Laden dieses Videos ist ein Problem aufgetreten. Dies könnte an einem vorübergehenden Netzwerkproblem liegen oder das Video ist möglicherweise nicht verfügbar.

Another cool stuff from Alibaba! MiniMax-H3 Acc LoRAs. Parallel Decoding Distillation for efficient video gen in a few steps. FL2VA + Ref2VA

Wildminder

11,471 subscribers

36,607 Aufrufe • vor 1 Monat •via X (Twitter)

Anya Rossi• Live Now

Private livecam show

3 Kommentare

Profilbild von ComfyUI Playground

ComfyUI Playgroundvor 1 Monat

Here is version for ComfyUI Enjoy!

Profilbild von 𝓝𝓲𝓬𝓴🇵🇳長月暑いね@猫の親バカ！！

𝓝𝓲𝓬𝓴🇵🇳長月暑いね@猫の親バカ！！vor 1 Monat

Is it compatible with ComfyUI?🤔

Profilbild von Diego P. Jaccottet

Diego P. Jaccottetvor 1 Monat

We need comparisons with other turbo LoRas.

Ähnliche Videos

MiniMax H3 🧐 offline prompt composer for video generation in ComfyUI. -for T2VA, I2VA, FL2VA, L2VA, and Ref2VA. -Reuse characters, environments, voices, and keyframes across scenes. -Custom Camera Builder, visual path planners, Advanced Editing github:

MiniMax H3 🧐 offline prompt composer for video generation in ComfyUI. -for T2VA, I2VA, FL2VA, L2VA, and Ref2VA. -Reuse characters, environments, voices, and keyframes across scenes. -Custom Camera Builder, visual path planners, Advanced Editing github:

Stable Diffusion Tutorials

18,877 Aufrufe • vor 1 Monat

MiniMax H3 😶‍🌫️ Sharpness LORA This supports ref2VA/fl2va, sharpens a source video while keeping it photorealistic. It conditions on the source through guide latents rather than through the model's native reference-video node. 👇

MiniMax H3 😶‍🌫️ Sharpness LORA This supports ref2VA/fl2va, sharpens a source video while keeping it photorealistic. It conditions on the source through guide latents rather than through the model's native reference-video node. 👇

Stable Diffusion Tutorials

76,249 Aufrufe • vor 17 Tagen

MiniMax Design made this chocolate ad look crazy smooth One prompt + a few minutes Pick from scene-ready Skills The motion is so clean Typography just works H3 really shines here ComfyUI runs locally Annual plans mean 20% off H3 + image gen

MiniMax Design made this chocolate ad look crazy smooth One prompt + a few minutes Pick from scene-ready Skills The motion is so clean Typography just works H3 really shines here ComfyUI runs locally Annual plans mean 20% off H3 + image gen

aiko

184,650 Aufrufe • vor 1 Monat

MiniMax Design makes this high energy animation PV stand out. H3 nails the overall look. Design has prompts, workflows, and Skills built in. One prompt is enough for the Agent to understand the idea and build it out. Done in just a few minutes. Annual users get 20% off H3 + image gen.

MiniMax Design makes this high energy animation PV stand out. H3 nails the overall look. Design has prompts, workflows, and Skills built in. One prompt is enough for the Agent to understand the idea and build it out. Done in just a few minutes. Annual users get 20% off H3 + image gen.

Rebecca Adson

82,083 Aufrufe • vor 1 Monat

Fast Byte Latent Transformer is accepted to ICML 2026! ⚡🥪 Byte-level LMs promise to free us from subword tokenizers, but decoding one byte at a time is super slow. We make BLT generation more efficient with BLT-D: text diffusion for parallel byte decoding. 1/

Fast Byte Latent Transformer is accepted to ICML 2026! ⚡🥪 Byte-level LMs promise to free us from subword tokenizers, but decoding one byte at a time is super slow. We make BLT generation more efficient with BLT-D: text diffusion for parallel byte decoding. 1/

Julie Kallini ✨

106,956 Aufrufe • vor 4 Monaten

MiniMax Design has Skills for all kinds of scenes Pick one + one line A few minutes later, it’s done ComfyUI for one click local deployment SD Director for precise visual control Annual: H3 + image gen 20% off This changes the workflow

MiniMax Design has Skills for all kinds of scenes Pick one + one line A few minutes later, it’s done ComfyUI for one click local deployment SD Director for precise visual control Annual: H3 + image gen 20% off This changes the workflow

Tyler Wayne

101,853 Aufrufe • vor 1 Monat

Tried creating a product animation with MiniMax H3 in MiniMax Design. The impressive part is how the Agent helps connect different stages — from concept planning to content creation. More than just generating a video.

Tried creating a product animation with MiniMax H3 in MiniMax Design. The impressive part is how the Agent helps connect different stages — from concept planning to content creation. More than just generating a video.

yuulia ✟

25,335 Aufrufe • vor 1 Monat

🎨 I trained 999 Style LoRAs with the fal Krea 2 Trainer. Every single image in this video is a different LoRA. That's 999 unique Style LoRAs, each trained for just 100 steps at 0.00035 LR, taking an average of ~30 seconds to train. As I said before, Krea 2 continues to prove that it's an incredible model for open-source Style LoRAs. Next week I'll be sharing even more LoRAs, along with a few surprises.

🎨 I trained 999 Style LoRAs with the fal Krea 2 Trainer. Every single image in this video is a different LoRA. That's 999 unique Style LoRAs, each trained for just 100 steps at 0.00035 LR, taking an average of ~30 seconds to train. As I said before, Krea 2 continues to prove that it's an incredible model for open-source Style LoRAs. Next week I'll be sharing even more LoRAs, along with a few surprises.

ilker

12,794 Aufrufe • vor 3 Monaten

MiniMax Design’s 3D Director is a lifesaver for a 3D newbie No wrestling with complex 3D tools Just keep chatting to create,storyboards and motion previews H3 prompts + a full workflow are built in Plus, annual users get 20% off both H3 and image gen

MiniMax Design’s 3D Director is a lifesaver for a 3D newbie No wrestling with complex 3D tools Just keep chatting to create,storyboards and motion previews H3 prompts + a full workflow are built in Plus, annual users get 20% off both H3 and image gen

Amalia

544,907 Aufrufe • vor 1 Monat

🎬Midjourney + Gen-3 + Hailuo AI + Udio: "Icepunk" (AI music video) Moodboards on Midjourney are powerful. I've uploaded only 10 images from my Icepunk interior collection to create this moodboard (--profile 6g973cs, feel free to try it!) and got a super cool and refreshing look for this music video - really helpful stuff! Images: Midjourney v 6.1 Video: Gen-3 and Hailuo AI (MiniMax) Music: Udio v 1.5 Editing: CapCut

Everett World

32,422 Aufrufe • vor 1 Jahr

#Viggle is back with an open model — Viggle-Animate — and it’s pretty damn capable. For the first time, they’re releasing weights + code. It replaces characters in video using MiniMax H3. Take a video, grab one frame, manually redraw the character as anything you want, and the model propagates that replacement through the entire clip while preserving motion, camera, and timing. No pose skeleton, segmentation, face tracking, depth, or text prompt. Just the video + one modified frame. Under the hood: a 33.1B MiniMax H3 ref2va fine-tune + DMD2 LoRA. Distillation brings it down to 4 sampling steps. It’s 6× faster than Wan2.2-Animate-14B. The catch: hardware. BF16 needs ~62 GiB for the transformer and peaks at ~80.1 GiB VRAM. They recommend 96 GB+ or CPU offload. It reportedly runs on a 32 GB RTX 5090 via NVFP4, but that quant isn’t public yet. Also, you can’t use a normal character portrait. The reference has to be a frame from the original video, already redrawn as the target character. #ai #web3

#Viggle is back with an open model — Viggle-Animate — and it’s pretty damn capable. For the first time, they’re releasing weights + code. It replaces characters in video using MiniMax H3. Take a video, grab one frame, manually redraw the character as anything you want, and the model propagates that replacement through the entire clip while preserving motion, camera, and timing. No pose skeleton, segmentation, face tracking, depth, or text prompt. Just the video + one modified frame. Under the hood: a 33.1B MiniMax H3 ref2va fine-tune + DMD2 LoRA. Distillation brings it down to 4 sampling steps. It’s 6× faster than Wan2.2-Animate-14B. The catch: hardware. BF16 needs ~62 GiB for the transformer and peaks at ~80.1 GiB VRAM. They recommend 96 GB+ or CPU offload. It reportedly runs on a 32 GB RTX 5090 via NVFP4, but that quant isn’t public yet. Also, you can’t use a normal character portrait. The reference has to be a frame from the original video, already redrawn as the target character. #ai #web3

VORTEX: AI Bros & AI Arena, Peak AI Buzz

47,030 Aufrufe • vor 21 Tagen

Let’s use Minimax H3 in ComfyUI in a more elegant way. Video tutorial for ComfyTV:@ image references and a dynamic number of image inputs, then videos edit & FX

Let’s use Minimax H3 in ComfyUI in a more elegant way. Video tutorial for ComfyTV:@ image references and a dynamic number of image inputs, then videos edit & FX

jtydhr88

14,680 Aufrufe • vor 1 Monat

I tried creating an educational video with MiniMax H3 in MiniMax Design. From structuring the idea to creating the visuals, the workflow helps simplify the entire process. It could be really useful for creators making tutorials and educational content.

I tried creating an educational video with MiniMax H3 in MiniMax Design. From structuring the idea to creating the visuals, the workflow helps simplify the entire process. It could be really useful for creators making tutorials and educational content.

Easton Coleford | AI

23,876 Aufrufe • vor 1 Monat

Another Chinese 'Sora': A new AI video tool launched today by Minimax, backed by major investors Alibaba Group and Tencent. 🎞️ Check out their official AI film Magic Coin🪙, created entirely with text-to-video . 🥁Try it for free now:

Another Chinese 'Sora': A new AI video tool launched today by Minimax, backed by major investors Alibaba Group and Tencent. 🎞️ Check out their official AI film Magic Coin🪙, created entirely with text-to-video . 🥁Try it for free now:

Junie Lau

661,140 Aufrufe • vor 2 Jahren

Bradley, the Gen Z Goliath from Kenya, steps into the ring for a local boxing match in Uganda!

Bradley, the Gen Z Goliath from Kenya, steps into the ring for a local boxing match in Uganda!

The Kenyan Vigilante

246,470 Aufrufe • vor 11 Monaten

I think I’m kinda obsessed with MiniMax Design now One prompt. That’s it. What was I even doing before this? Just a few minutes and one prompt gets you a full retro Japanese travel-poster-style video. MiniMax Design really brings out the best in H3 And no, I’m not exaggerating

I think I’m kinda obsessed with MiniMax Design now One prompt. That’s it. What was I even doing before this? Just a few minutes and one prompt gets you a full retro Japanese travel-poster-style video. MiniMax Design really brings out the best in H3 And no, I’m not exaggerating

Jaynit Makwana

36,397 Aufrufe • vor 1 Monat

Presto! Distilling Steps and Layers for Accelerating Music Generation Despite advances in diffusion-based text-to-music (TTM) methods, efficient, high-quality generation remains a challenge. We introduce Presto!, an approach to inference acceleration for score-based diffusion transformers via reducing both sampling steps and cost per step. To reduce steps, we develop a new score-based distribution matching distillation (DMD) method for the EDM-family of diffusion models, the first GAN-based distillation method for TTM. To reduce the cost per step, we develop a simple, but powerful improvement to a recent layer distillation method that improves learning via better preserving hidden state variance. Finally, we combine our step and layer distillation methods together for a dual-faceted approach. We evaluate our step and layer distillation methods independently and show each yield best-in-class performance. Our combined distillation method can generate high-quality outputs with improved diversity, accelerating our base model by 10-18x (230/435ms latency for 32 second mono/stereo 44.1kHz, 15x faster than comparable SOTA) -- the fastest high-quality TTM to our knowledge.

Presto! Distilling Steps and Layers for Accelerating Music Generation Despite advances in diffusion-based text-to-music (TTM) methods, efficient, high-quality generation remains a challenge. We introduce Presto!, an approach to inference acceleration for score-based diffusion transformers via reducing both sampling steps and cost per step. To reduce steps, we develop a new score-based distribution matching distillation (DMD) method for the EDM-family of diffusion models, the first GAN-based distillation method for TTM. To reduce the cost per step, we develop a simple, but powerful improvement to a recent layer distillation method that improves learning via better preserving hidden state variance. Finally, we combine our step and layer distillation methods together for a dual-faceted approach. We evaluate our step and layer distillation methods independently and show each yield best-in-class performance. Our combined distillation method can generate high-quality outputs with improved diversity, accelerating our base model by 10-18x (230/435ms latency for 32 second mono/stereo 44.1kHz, 15x faster than comparable SOTA) -- the fastest high-quality TTM to our knowledge.

AK

30,446 Aufrufe • vor 2 Jahren

MiniMax H3 🧐 Comfy Tested -Minimum requirements for 480p video on this model is a 3060 with 12GB vram, 32GB of system ram and a good nvme SSD. -generating a 5 second (124 frames) 480p (864x480) video on this system and it took a bit less than 9 minutes end to end (20 steps)

MiniMax H3 🧐 Comfy Tested -Minimum requirements for 480p video on this model is a 3060 with 12GB vram, 32GB of system ram and a good nvme SSD. -generating a 5 second (124 frames) 480p (864x480) video on this system and it took a bit less than 9 minutes end to end (20 steps)

Stable Diffusion Tutorials

102,325 Aufrufe • vor 1 Monat

VideoDeltaNet - near-lossless hybrid attention on MiniMax H3. You can now generate a 14s video faster than you can watch it. - 14.4s 768p clip in 11.23s on 8 B200s - local softmax for detail + linear for long range - streams H3-level video continuously - open weights + inference stack -tops FastH3 quality & instruction-following

VideoDeltaNet - near-lossless hybrid attention on MiniMax H3. You can now generate a 14s video faster than you can watch it. - 14.4s 768p clip in 11.23s on 8 B200s - local softmax for detail + linear for long range - streams H3-level video continuously - open weights + inference stack -tops FastH3 quality & instruction-following

Wildminder

20,821 Aufrufe • vor 26 Tagen

AI video isn’t expensive. Bad prompting is. I built a JSON template for MiniMax H3 so you just drop your idea in and hit generate. 5 different use cases below, tested on magnific 👇

AI video isn’t expensive. Bad prompting is. I built a JSON template for MiniMax H3 so you just drop your idea in and hit generate. 5 different use cases below, tested on magnific 👇

TechHalla

59,326 Aufrufe • vor 1 Monat