Wildminder's banner

Wildminder

@wildmindai • 10,137 subscribers

Physicist, Programmer, Designer

Shorts

Another cool stuff from NVIDIA. LocateAnything - high-speed visual search engine. You provide a text prompt and it instantly pinpoints that object's exact location in an image. - 10x speedup for dense object detection - Qwen2.5-3B + Moon-ViT - Fast/Slow/Hybrid modes - trained on 138M samples for UI, docs, generic grounding.

Another cool stuff from NVIDIA. LocateAnything - high-speed visual search engine. You provide a text prompt and it instantly pinpoints that object's exact location in an image. - 10x speedup for dense object detection - Qwen2.5-3B + Moon-ViT - Fast/Slow/Hybrid modes - trained on 138M samples for UI, docs, generic grounding.

50,143 次观看

NVIDIA says: no more "brute force every pixel" of video understanding. AutoGaze- identifies and removes redundant video patches before they enter a Vision Transformer. Now we can processes 4K long-video in real-time. Works with SigLIP2 and NVILA.

NVIDIA says: no more "brute force every pixel" of video understanding. AutoGaze- identifies and removes redundant video patches before they enter a Vision Transformer. Now we can processes 4K long-video in real-time. Works with SigLIP2 and NVILA.

296,543 次观看

CogOmniControl by Tencent. Reasoning-driven controllable video gen. CogVLM + CogOmniDiT to translate sparse storyboards/sketches into production-quality video. beats VINO, VACE-Wan2.1

CogOmniControl by Tencent. Reasoning-driven controllable video gen. CogVLM + CogOmniDiT to translate sparse storyboards/sketches into production-quality video. beats VINO, VACE-Wan2.1

23,795 次观看

LTX2.3 ReStyle LoRA Transfers simpler styles (flat 2D, cel-shaded, monochrome line art). struggles with complex styles (texture, intricate detail, strong material/lighting effects).

LTX2.3 ReStyle LoRA Transfers simpler styles (flat 2D, cel-shaded, monochrome line art). struggles with complex styles (texture, intricate detail, strong material/lighting effects).

24,442 次观看

LGTM from Apple: 4K feed-forward 3D Gaussian Splatting. instant 4K 3D scenes without massive GPUs.. - predicts a few lightweight 3D shapes, wraps them in ultra-high-res 2D textures. - low Memory usage You take two normal photos of a room. Instantly walk around it in flawless 3D.

LGTM from Apple: 4K feed-forward 3D Gaussian Splatting. instant 4K 3D scenes without massive GPUs.. - predicts a few lightweight 3D shapes, wraps them in ultra-high-res 2D textures. - low Memory usage You take two normal photos of a room. Instantly walk around it in flawless 3D.

46,278 次观看

Video diffusion models are just overqualified depth estimators! Deterministic single-pass depth estimation based on WanV2.1. - SOTA 5.5 AbsRel on ScanNet - data-efficient than baselines; - no temporal flicker + infinite-length estimation w/ zero scale drift.

Video diffusion models are just overqualified depth estimators! Deterministic single-pass depth estimation based on WanV2.1. - SOTA 5.5 AbsRel on ScanNet - data-efficient than baselines; - no temporal flicker + infinite-length estimation w/ zero scale drift.

49,209 次观看

LTX 2.3 Creative Upscale IC-LoRA. - Generative second-pass refiner for soft or low-resolution video; - enhances detail and clarity without standard upscaling; - output varies based on workflow/settings.

LTX 2.3 Creative Upscale IC-LoRA. - Generative second-pass refiner for soft or low-resolution video; - enhances detail and clarity without standard upscaling; - output varies based on workflow/settings.

16,526 次观看

3D modeling entirely replaced by stick figures. SK-Adapter brings skeleton-based structural control for native 3D. > Feed it a basic skeleton > Type what you want to see > Get a fully rendered 3D character in under 15 seconds > Already rigged and ready for animation > Zero Blender experience required Game devs are happy.

3D modeling entirely replaced by stick figures. SK-Adapter brings skeleton-based structural control for native 3D. > Feed it a basic skeleton > Type what you want to see > Get a fully rendered 3D character in under 15 seconds > Already rigged and ready for animation > Zero Blender experience required Game devs are happy.

26,931 次观看

Unsloth dropped new LTX-2.3 GGUFs. > Dev/distilled UD-Q2/Q5

Unsloth dropped new LTX-2.3 GGUFs. > Dev/distilled UD-Q2/Q5

24,502 次观看

ComfyUI-WanVideoWrapper now supports SteadyDancer: like WanAnimate - human image animation framework; produces high-fidelity, coherent motion

ComfyUI-WanVideoWrapper now supports SteadyDancer: like WanAnimate - human image animation framework; produces high-fidelity, coherent motion

42,246 次观看

Thanks to Kijai, One-to-All Animation has already been added to ComfyUI.

Thanks to Kijai, One-to-All Animation has already been added to ComfyUI.

33,914 次观看

LightVAE + ComfyUI node: High-performance video VAE; runs 2–3x faster using 50% less memory; LightTAE offers a 10+x speedup on just ~0.4GB VRAM

LightVAE + ComfyUI node: High-performance video VAE; runs 2–3x faster using 50% less memory; LightTAE offers a 10+x speedup on just ~0.4GB VRAM

38,092 次观看

As usual, Kijai has prepared the Wan-Move, and it is available in ComfyUI.

As usual, Kijai has prepared the Wan-Move, and it is available in ComfyUI.

25,558 次观看

Capybara? 14B model for T2V, T2I, TV2V, TI2I. - based on HunyuanVideo1.5; - byt5-small, Glyph-SDXL-v2, SigLIP; - 480p-1080p; 16.7GB model, 5GB VAE.. mostly for video editing.

Capybara? 14B model for T2V, T2I, TV2V, TI2I. - based on HunyuanVideo1.5; - byt5-small, Glyph-SDXL-v2, SigLIP; - 480p-1080p; 16.7GB model, 5GB VAE.. mostly for video editing.

16,853 次观看

AnyDepth: Lightweight zero-shot monocular depth estimation; surpasses DPT; - nicely preserves detail.

AnyDepth: Lightweight zero-shot monocular depth estimation; surpasses DPT; - nicely preserves detail.

18,152 次观看

Your LTX-2 performance boost has arrived. NVIDIA Studio Driver (591.74 January)- optimizations for LTX2 + support for NVFP4/NVFP8 in ComfyUI.

Your LTX-2 performance boost has arrived. NVIDIA Studio Driver (591.74 January)- optimizations for LTX2 + support for NVFP4/NVFP8 in ComfyUI.

13,803 次观看

Videos

Anya Rossi

sweetdream.ai

SweetDream.ai•Sponsored•Livecam

Watch Anya Live

Anya is streaming live right now! Join her private show and enjoy exclusive content.

Exclusive private shows

1.2k viewers online

Private Show

Join now for exclusive access

Free preview available • Premium content

Netflix dropped some useful stuff. VOID -video object and interaction deletion. - removes objects while realistically simulating physical consequences; - beats Runway/ProPainter; - CogVideoX-5B + SAM 2; looks good, no smudges/artifacts

Netflix dropped some useful stuff. VOID -video object and interaction deletion. - removes objects while realistically simulating physical consequences; - beats Runway/ProPainter; - CogVideoX-5B + SAM 2; looks good, no smudges/artifacts

365,140 次观看 • 2 个月前

Another cool stuff from NVIDIA. LocateAnything - high-speed visual search engine. You provide a text prompt and it instantly pinpoints that object's exact location in an image. - 10x speedup for dense object detection - Qwen2.5-3B + Moon-ViT - Fast/Slow/Hybrid modes - trained on 138M samples for UI, docs, generic grounding.

Another cool stuff from NVIDIA. LocateAnything - high-speed visual search engine. You provide a text prompt and it instantly pinpoints that object's exact location in an image. - 10x speedup for dense object detection - Qwen2.5-3B + Moon-ViT - Fast/Slow/Hybrid modes - trained on 138M samples for UI, docs, generic grounding.

50,143 次观看 • 8 天前

llmfit. Useful tool that probes hardware and tells you exactly which LLMs will actually run. - handles MoE expert offloading, picks the best quantization for your RAM, estimates tokens/sec before you even pull the weights. Essential for local dev.

llmfit. Useful tool that probes hardware and tells you exactly which LLMs will actually run. - handles MoE expert offloading, picks the best quantization for your RAM, estimates tokens/sec before you even pull the weights. Essential for local dev.

246,976 次观看 • 3 个月前

Awesome. NVIDIA dropped PiD - fast high-res latent decoding via pixel diffusion! - replace VAE - 4/8x upsampling - 2k decoding in <1s on RTX 5090 - works with FLUX.1/SD3/Z - rapid generation previews sharper details, much lower hardware lag compared to standard methods.

Awesome. NVIDIA dropped PiD - fast high-res latent decoding via pixel diffusion! - replace VAE - 4/8x upsampling - 2k decoding in <1s on RTX 5090 - works with FLUX.1/SD3/Z - rapid generation previews sharper details, much lower hardware lag compared to standard methods.

23,205 次观看 • 10 天前

MoZoo- Wan2.1 for CGI animals with fur and muscle simulation. - ADA + RAR-RoPE for motion alignment and texture transfer - without manual rigging/hair-sim - beats VACE in temporal consistency

MoZoo- Wan2.1 for CGI animals with fur and muscle simulation. - ADA + RAR-RoPE for motion alignment and texture transfer - without manual rigging/hair-sim - beats VACE in temporal consistency

13,628 次观看 • 6 天前

LTX2.3 Obscura Remove LoRA. Some kind of a "digital X-ray" for video cleanup. - Strips foreground junk. Peel back smoke, haze, or clutter - Reconstructs hidden scenes. Describe the background to fill holes.

LTX2.3 Obscura Remove LoRA. Some kind of a "digital X-ray" for video cleanup. - Strips foreground junk. Peel back smoke, haze, or clutter - Reconstructs hidden scenes. Describe the background to fill holes.

54,849 次观看 • 1 个月前

A totally new level of pose control in ComfyUI - VNCCS. - full 3D Pose Studio for character posing & lighting, - multi-pose, body generator, pose gallery; - vision-guided QWEN Detailer + camera controls.

A totally new level of pose control in ComfyUI - VNCCS. - full 3D Pose Studio for character posing & lighting, - multi-pose, body generator, pose gallery; - vision-guided QWEN Detailer + camera controls.

163,005 次观看 • 4 个月前

PanoWorld. An interesting way to use Qwen-Edit. It converts 2D floor plans into photorealistic, consistent VR home tours. Great for real estate and interior designers. It lets you walk through a home that hasn’t been built or furnished yet. Ensures seamless 360 views via CPRoPE

PanoWorld. An interesting way to use Qwen-Edit. It converts 2D floor plans into photorealistic, consistent VR home tours. Great for real estate and interior designers. It lets you walk through a home that hasn’t been built or furnished yet. Ensures seamless 360 views via CPRoPE

20,103 次观看 • 14 天前

FLUX.2 Klein in pure pixel space! No VAE. AsymFlow - hyper-realistic images by working directly in pixel space rather than using compressed latent representations. - sharper textures, superior visual fidelity - 40% faster - low-rank noise parameterization to solve high-dimensional bottlenecks ComfyUI support incoming

FLUX.2 Klein in pure pixel space! No VAE. AsymFlow - hyper-realistic images by working directly in pixel space rather than using compressed latent representations. - sharper textures, superior visual fidelity - 40% faster - low-rank noise parameterization to solve high-dimensional bottlenecks ComfyUI support incoming

26,976 次观看 • 21 天前

CS:GO + Adobe + Wan2.1 = WorldCam. Interactive autoregressive 3D gaming worlds. > AI now generates playable 3D worlds live > You move the mouse, the AI builds the map instantly > Turn around, and it remembers exactly what was there No traditional game engines. Just neural networks hallucinating reality.

CS:GO + Adobe + Wan2.1 = WorldCam. Interactive autoregressive 3D gaming worlds. > AI now generates playable 3D worlds live > You move the mouse, the AI builds the map instantly > Turn around, and it remembers exactly what was there No traditional game engines. Just neural networks hallucinating reality.

67,670 次观看 • 2 个月前

HappyHorse-1.0 vs All. - 720p, 24fps - crisp - pleasant colors, no AI-ish vibrancy - nice textures, good skin - dynamic - good prompt following If this model runs on consumer GPUs - it'll be an absolute nuke

HappyHorse-1.0 vs All. - 720p, 24fps - crisp - pleasant colors, no AI-ish vibrancy - nice textures, good skin - dynamic - good prompt following If this model runs on consumer GPUs - it'll be an absolute nuke

49,828 次观看 • 1 个月前

Pixal3D by Tencent. 3D models from images with unprecedented accuracy. - Explicit 2D-3D correspondence - Aggregates multi-view feature volumes for global consistency - 93.57% IoU on Toys4K benchmark - Outperforms TRELLIS/Hunyuan3D-2.1 based on Direct3D-S2 + DINOv2

Pixal3D by Tencent. 3D models from images with unprecedented accuracy. - Explicit 2D-3D correspondence - Aggregates multi-view feature volumes for global consistency - 93.57% IoU on Toys4K benchmark - Outperforms TRELLIS/Hunyuan3D-2.1 based on Direct3D-S2 + DINOv2

21,766 次观看 • 23 天前

Sweet! Kijai has added standalone ComfyUI nodes for SCAIL pose processing.

Sweet! Kijai has added standalone ComfyUI nodes for SCAIL pose processing.

119,435 次观看 • 5 个月前

Soprano: An instant, ultra-lightweight TTS model for realistic speech; generates 10 hours of 32kHz audio in <20s; streams with <15ms latency using just 80M params & <1GB VRAM. Has some limitations and drawbacks.

Soprano: An instant, ultra-lightweight TTS model for realistic speech; generates 10 hours of 32kHz audio in <20s; streams with <15ms latency using just 80M params & <1GB VRAM. Has some limitations and drawbacks.

111,168 次观看 • 5 个月前

BiRefNet background removal now native in ComfyUI, no custom nodes needed Actually fast and clean.

BiRefNet background removal now native in ComfyUI, no custom nodes needed Actually fast and clean.

22,618 次观看 • 26 天前

LTX is cooking hard. Just-Dub-It IC LoRA, and you can dub anything. - Instead of separate steps for voice and lips, it generates them simultaneously - handles occlusions and profile views - beats HeyGen/MuseTalk - maintains bg noise and speaker identity also it stretches the video and audio together so the pacing matches the new language perfectly. Just awesome!

19,732 次观看 • 24 天前

Good. ComfyUI just got native MoGe support - 3D geometry from monocular images.

Good. ComfyUI just got native MoGe support - 3D geometry from monocular images.

15,988 次观看 • 21 天前

No way. Adobe's new node editor is a straight-up ComfyUI clone! The killer feature is encapsulating an entire graph into a simple tool that can be dropped into Photoshop.

No way. Adobe's new node editor is a straight-up ComfyUI clone! The killer feature is encapsulating an entire graph into a simple tool that can be dropped into Photoshop.

110,418 次观看 • 7 个月前

NVIDIA just dropped UniRelight, a handy video-relighting model - estimates albedo, relit video in a single pass - based on Cosmos-Predict1-7B - supports complex materials - temporal consistency - beats DiLightNet, NeuralGaffer

NVIDIA just dropped UniRelight, a handy video-relighting model - estimates albedo, relit video in a single pass - based on Cosmos-Predict1-7B - supports complex materials - temporal consistency - beats DiLightNet, NeuralGaffer

29,374 次观看 • 1 个月前

Awesome! OmniVoice-TTS in ComfyUI. - zero-shot multilingual TTS; - 600+ languages; - voice cloning; - voice design; - multi-speaker dialogues; - supports SageAttention and non-verbal expression tags.

Awesome! OmniVoice-TTS in ComfyUI. - zero-shot multilingual TTS; - 600+ languages; - voice cloning; - voice design; - multi-speaker dialogues; - supports SageAttention and non-verbal expression tags.

35,799 次观看 • 2 个月前