Üretim4
görselvideo·reddit-stablediffusion·10 saat önce
Repo here- https://github.com/Adudeguyman/ComfyUI-Fantastic-MiniMaxH3-PromptBuilder Ooookay back again. I made RefMods work a bit better and solved a lot of identity/bleed issues. At the cost of time and memory, of course. What it do- OG vanilla refmods would save your media as latents in a .safetensors file, and feed all of those into the MiniMax H3 DiT directly, with no references and the text encoder never saw it. This worked fine with one refmod since the model knew it had to do SOMETHING with those references, and it did it surprisingly well. But that also lead to people stacking…
video·hf-papers·2 gün önce·16 oy
High-resolution video generation is expensive, as its cost grows rapidly with the number of spatiotemporal tokens. A practical alternative first generates a lower-resolution video and then applies a refiner, but conventional multi-step refinement introduces a second sampling bottleneck. We present SoL-Refiner, a one-step video refiner that transforms low-resolution model outputs into 4K videos with a single denoising step. Our three-stage recipe combines high-resolution continual training, reinforcement learning (RL) post-training, and a final one-step distillation. We introduce…
video·arxiv-cv·19 saat önce·+1 kaynak
Teaching humanoids loco-manipulation skills, such as carrying diverse objects, via visual imitation is a promising path toward generalist robots. However, collecting diverse, high-quality interaction videos, such as clips that clearly show a person's full body and unoccluded interactions with objects, poses a practical barrier to scaling this approach. We propose PRISM, a real-to-sim-to-real framework that overcomes this limitation by amplifying a handful of real videos into a large, diverse training set. PRISM first generates hundreds of diverse "counterfactual" human-object interaction…
görsel·reddit-stablediffusion·7 saat önce
A LoRA designed to reduce the typical smooth/plastic AI look and bring natural texture and realism to your images. What's new in V5 (KL9_V5_Pro): Summary: This version gives the best natural texture so far. V5 is trained on a high-quality dataset of SFW professional photography with a wide variety of subjects: portraits, lifestyle, fashion, wildlife, sports and fitness, architecture, nature, and more. This makes it far more versatile than earlier versions, which focused mainly on skin. You now get realistic detail, texture, and lighting across all kinds of scenes, not just close-ups of…
Oyun2
oyun·reddit-gamedev·az önce
Hi all, I am developing a game for fun using my own custom game engine (C++ and OpenGL). I have been working on this game (and the engine) for 4 years now. I'm a French software engineer working for an American company, so this project is not a full-time job. I can only work on it during my free time after work or on weekends. I am not doing this for money, I just want people to play my game and enjoy it when it's ready. I think I still have about one year of development left to get a fully playable game that I can share. The engine and performances are now in a very good state. What I really…
oyun·reddit-gamedev·1 saat önce
So curated events (seasonal, by publisher, by region, by genre....) were a staple of the Steam ecosystem for a long time. I can personally vouch for their effectiveness in giving exposure to smaller indie titles if they managed to get featured in an event. This seems like a massive blow to indies discoverability. Any more information? What do you think? https://steamcommunity.com/groups/steamworks/announcements/detail/676258795703240952 EDIT: Okay, after reading the discussion below, as well as the discussion on steam discussion forums, it is clear that Valve is not clear as to what this will…
Modeller & Araştırma4
↑ yükselişte53 oy/saataraçmodelşirket·simonwillison·19 saat önce·987 oy·+1 kaynak
My comment on GPT 6.1 Sol: Near-Astra intelligence for a fifth of the price — Hacker News. I'm a bit late with the pelicans because I was live-blogging the keynote: https://simonwillison.net/2026/Sep/29/openai-devday-2026-liv... Here they are for GPT-6.1-Sol: https://tools.simonwillison.net/markdown-svg-renderer?url=ht... They're not notably different from the GPT-6 family pelicans: https://static.simonwillison.net/static/2026/gpt-pelicans-gr... Tags: ai , openai , generative-ai , llms , pelican-riding-a-bicycle , gpt
↑ yükselişte33 oy/saatmodel·openai·2 gün önce
Dots by OpenAI are proactive assistants that can keep working across complex projects and everyday tasks. Learn how dots help you stay in control while work moves forward.
↑ yükselişteHN'de GPT-6.1 0→7 hikâyemodel·openai·1 gün önce
Meet GPT-6.1 Sol: near-Astra intelligence for coding, computer use, and professional work at one-fifth of Astra’s standard API input and output token prices.
↑ yükselişte27 oy/saataraçmodelşirket·hackernews·21 saat önce·545 oy