File: K2-00036-Kamen Rider tv a(...).png (1.59 MB, 768x1152)

1.59 MB PNG

Afternoon anons
>>109792036
Awesome OP

File: K2-00037-Kamen Rider tv a(...).png (1.68 MB, 768x1152)

1.68 MB PNG

Happy 9/11

>ga

File: 00009-1226468051.jpg (586 KB, 2016x2592)

586 KB JPG

>>109792296
afternoon

>mfw Resource news

09/11/2026

>ComfyUI-VDN-H3-24GB
https://github.com/Speach1sdef178/ComfyUI-VDN-H3-24GB

>LTX-2.5 22B IC-LoRA Reference Sheet Control
https://huggingface.co/Lightricks/LTX-2.5-22b-IC-LoRA-Ingredients#ltx-25-22b-ic-lora-reference-sheet-control

>FastH3-Live v1.2.0 update
https://huggingface.co/datasets/jacokon/fasth3-live

>MiniMax-H3-Image-Training-Adapter
https://huggingface.co/circlestone-labs/MiniMax-H3-Image-Training-Adapter

>TaoMate-H3: Low-latency streaming audio-video generation runtime built on MiniMax H3
https://github.com/TaoLiveAIGC/TaoMate-H3

>FrameForge Motion Context Video Editor for ComfyUIR
https://github.com/spacesimeco-hue/Chain-Motion-AI-Video-Editor

>From Evaluation to Enhancement: Benchmarking and Improving Think-with-Video Reasoning for Video Generative Models
https://huggingface.co/datasets/KlingTeam/VWG-Bench

>Uncertainty DMD: Restoring Diversity in Few-Step Autoregressive Video Distillation
https://scdzx.github.io/Uncertainty-DMD

>A Multi-View and Confusion-Guided Ensemble Framework for Robust Synthetic Image Attribution
https://github.com/ZOMIN28/SIA

>FreeFlow: A Bias-free Hierarchical Transformer for Optical Flow Estimation
https://github.com/msu-video-group/freeflow

>World in World: Explore the World with World Models
https://chenxi-song.github.io/worldinworld

>OmniKVQuant: KV Cache Quantization for Omni-LLMs
https://github.com/kaistmm/OmniKVQuant

>BSAI-MiniMAX-H3-Prompt
https://github.com/xm6018924/BSAI-MiniMAX-H3-Prompt

>ComfyUI Native Yue2
https://huggingface.co/Comfy-Org/Yue2

09/10/2026

>Interpreting Object-Dependent Concept Brittleness in Text-to-Image Diffusion Models
https://github.com/Metecade/Object-Dependent-Concept-Brittleness

>Mask Forcing: Improving Autoregressive Video Diffusion Distillation via Dual-Noise Masking Rollout
https://alicezrzhao.github.io/mask_forcing

>Geodesic-informed Generative Diffusion Model For Topology-preserved Image Video Generation
https://github.com/nellie689/IGG

>mfw Research news

09/11/2026

>Harnessing Intrinsic Subject-Aware Attention for Controllable Multi-Subject Video Generation
https://arxiv.org/abs/2609.11507

>AcFlow: Controlling Text-to-Image Diffusion Transformers via Learned Conditional Activation Flow
https://arxiv.org/abs/2609.10723

>Vidu S2: Real-Time Interactive, Editable, and Spatial Video Generation
https://arxiv.org/abs/2609.11638

>Overpainting: Localized Context-aware Diffusion Image Editing
https://overpainting.github.io

>Learning Interaction between Image and Layout Priors for Joint Image-Layout Generation in Design Templates
https://arxiv.org/abs/2609.11519

>OmniHallu: Unified Hallucination Detection for Cross-Modal Comprehension and Generation in Multimodal Large Language Models
https://arxiv.org/abs/2609.11244

>Shedding Light: A Benchmark for Evaluating Lighting Understanding in Generative Image Models
https://lvsn.github.io/SheddingLight

>Logit Refiner: Improving Visual Autoregressive Models via Intra-Scale Dependency Modeling
https://compvis.github.io/logit-refiner

>CamPilot: A Multi-Agent Cinematic Assistant for Camera-Controlled Movie Generation
https://arxiv.org/abs/2609.10943

>Mi-Ripple: Restoring Images Degraded by Iterative AI Editing
https://arxiv.org/abs/2609.11317

>SenseNova-U1.5: Towards Native Unified Visual Intelligence
https://arxiv.org/abs/2609.11929

>LoopVAE: Recurrent Depth Across Scales for Visual Tokenization
https://arxiv.org/abs/2609.11516

>Revisiting Avatar-As-Image: High-Fidelity Registration is All You Need
https://yuxuan-xue.com/avaimg

>Are We Really Doing Few-Shot Learning? A Critical Examination of Pre-Training Assumptions
https://arxiv.org/abs/2609.10851

>MultiHuSE: A Multimodal Dataset for Humour Styles and Emotions
https://arxiv.org/abs/2609.11322

>Leveraging Avatar Fingerprinting: A Multi-Generator Photorealistic Talking-Head Public Database and Benchmark
https://arxiv.org/abs/2603.26934

File: debo_dcl_k2_00029_.png (2.81 MB, 1664x1069)

2.81 MB PNG

>>109792301
idk why but this post cracks me up

File: 00332-863979754310866-kro(...).jpg (2.45 MB, 3072x2304)

2.45 MB JPG

>>109794811
it's a watercolor kind of day

>>109792301
Happy belated 9/11 <3

>>109792938
>>109793235
>>109793464
>>109794698
I really like these ones, debo. Style prompts pl0x?

File: debo_dcl_k2_00037_.png (2.53 MB, 1664x1069)

2.53 MB PNG

>>109794959
glad you like them! the style is very wildcarded, like usual
https://files.catbox.moe/2f32ig.png

{oil painting|watercolor|gouache|acrylic|fresco|tempera|ink wash|pastel|charcoal|mixed media|woodblock print|lithograph|engraving|etching|collage|screen print|propaganda poster|newsprint illustration|war photography|documentary illustration} rendered in {cubist geometric fragmentation|expressionist emotional distortion|brutalist graphic severity|gritty realist detail|dark romanticism|social realism|german expressionism|constructivist stark geometry|propaganda poster boldness|film noir shadow|dystopian illustration style|industrial age engraving|wartime documentary style|soviet realist painting|ashcan school grit|neue sachlichkeit cold precision|kitchen sink realism|raw outsider intensity|politically charged symbolism|bleak naturalism}

I think the dystopian theme is influencing the style as well

>>109795028
Wew, thanks, looks like I now gotta figure out which is which though.

File: debo_dcl_k2_00038_.png (2.59 MB, 1664x1069)

2.59 MB PNG

>>109795065
the catbox shows the final output too. that one was
>oil painting rendered in neue sachlichkeit cold precision

>>109795077
>the catbox shows the final output too. that one was
Yeah but that was one out of four. But nevermind, debo, I should probably just try all of these out (or at least the ones I hadn't already before) anyhow.

File: debo_dcl_k2_00041_.png (2.49 MB, 1664x1069)

2.49 MB PNG

>>109795163
ah, you were hoping for each. np
31: https://files.catbox.moe/ygvu00.png
33: https://files.catbox.moe/3tgepd.png
34: https://files.catbox.moe/t8c006.png

have fun :)

File: debo_dcl_k2_00042_.png (3.13 MB, 1664x1069)

3.13 MB PNG

>>109795376
shes been very possessed lately

File: 00358-975117902233194-kro(...).jpg (2.17 MB, 3072x2304)

2.17 MB JPG

>>109795399
havent we all

File: 00361-397223361661690-kro(...).jpg (1.86 MB, 3072x2304)

1.86 MB JPG

me on the left

File: 00373-851659000189681-kro(...).jpg (2 MB, 3072x2304)

2 MB JPG

gn all

File: debo_dcl_k2_00048_.png (2.75 MB, 1664x1069)

2.75 MB PNG

>>109795726
gn

File: sekhmet-krea2t-2026-09-10(...).png (2.96 MB, 1920x1080)

2.96 MB PNG

>>109795726
see ya

File: sketchy-krea2t-2026-09-12(...).png (3.53 MB, 1920x1080)

3.53 MB PNG

goddamnit they're taking 2 minutes again. but i don't wanna restart!!!

File: debo_dcl_k2_00049_.png (3.4 MB, 1664x1069)

3.4 MB PNG

>>109795962
you have to defragment your tensors

File: poly64-krea2t-2026-09-12_(...).png (2.37 MB, 1920x1080)

2.37 MB PNG

>>109795994
aka restart windows lel

File: .gif (500 KB, 500x318)

500 KB GIF

too busy throwing crap at the free-for-all free v6 giveaway wall

File: debo_dcl_k2_00050_.png (3.5 MB, 1664x1069)

3.5 MB PNG

>>109796042
I tried a handful of ideas but didn't get anything that stuck
I think variety slider is a bait

File: poly64-krea2t-2026-09-12_(...).png (2.27 MB, 1920x1080)

2.27 MB PNG

>>109796079
yeah it's trash. i already told you. or i remember telling you who knows

File: debo_dcl_k2_00051_.png (3.13 MB, 1664x1069)

3.13 MB PNG

>>109796083
I had to find out for myself
but now that leaves us with the mystery: how to make v6 work

File: poly64-krea2t-2026-09-12_(...).png (2.13 MB, 1920x1080)

2.13 MB PNG

>>109796079
oh and got an updated style prompt skill thingy, it's OK. better than nothing
https://gist.github.com/illuminatianon/1ed3b3bed7fde93910d2efa17ddb1eb7
astra made it so it must be good. it's called a skill but i use it in a GPT project as a source and it's FINE

File: debo_dcl_k2_00054_.png (2.77 MB, 1664x1069)

2.77 MB PNG

>>109796100
ritual is dead
long live ritual v.astra

File: poly64-krea2t-2026-09-12_(...).png (2.04 MB, 1920x1080)

2.04 MB PNG

>>109796119
ritual is for lyric writing, this the engine for style prompts and kind of lyrics, just don't ask for lyrics after a style bc it'll write a song about the style lol