I built a site to create free AI videos using LTX 2.3 running on my own GPUs
https://redd.it/1t9juoy
@rStableDiffusion
https://redd.it/1t9juoy
@rStableDiffusion
This media is not supported in your browser
VIEW IN TELEGRAM
TenStrip's Workflow is the first LTX 2.3 workflow I found that actually works for Spicy Content it's almost like using the old Grok.
https://redd.it/1t9pbjd
@rStableDiffusion
https://redd.it/1t9pbjd
@rStableDiffusion
The Anima realism model is crazy good. Don’t miss it!
I’ve been messing with the anima realism model posted here (https://civitai.red/models/2585622/ultrareal-fine-tune-anima). If you want prompt adherence for weird stuff, it does a really good job. What’s cool is you can do hybrid danbooru / natural language and it just goes with it.
I’m stunned at how good it is and surprised it’s not getting more traction, especially since this is the authors experiment and the model and this finetune aren’t done yet. The output is decent if you prompt well. It’s not as photo realistic as ZIT or whatever but it will do all your weird danbooru tags other ones blush over. I actually think for the amateur photography all you guys want here it’s a good model.
I do 50 steps , 5cfg, euler (not ancestral). Anima is slow as hell on my Mac for such a small model but hoping the devs improve it somehow. It also works with the turbo lora!
Additionally I saw someone extracted the realism ‘stuff’ as a lora. It’s in the comments of the civitai page, linked in a random Google Drive.
Anyway try it out and if the author sees this thanks dude. Lmk if I can chip in for another training run. There is so much potential here.
https://redd.it/1t9r8c6
@rStableDiffusion
I’ve been messing with the anima realism model posted here (https://civitai.red/models/2585622/ultrareal-fine-tune-anima). If you want prompt adherence for weird stuff, it does a really good job. What’s cool is you can do hybrid danbooru / natural language and it just goes with it.
I’m stunned at how good it is and surprised it’s not getting more traction, especially since this is the authors experiment and the model and this finetune aren’t done yet. The output is decent if you prompt well. It’s not as photo realistic as ZIT or whatever but it will do all your weird danbooru tags other ones blush over. I actually think for the amateur photography all you guys want here it’s a good model.
I do 50 steps , 5cfg, euler (not ancestral). Anima is slow as hell on my Mac for such a small model but hoping the devs improve it somehow. It also works with the turbo lora!
Additionally I saw someone extracted the realism ‘stuff’ as a lora. It’s in the comments of the civitai page, linked in a random Google Drive.
Anyway try it out and if the author sees this thanks dude. Lmk if I can chip in for another training run. There is so much potential here.
https://redd.it/1t9r8c6
@rStableDiffusion
civitai.red
UltraReal Fine-Tune Anima - v1.2 Anima_Preview3 | Anima Checkpoint | Civitai
--preview3 I'm also sharing a new experimental version of UltraReal FineTune Anima, this time trained on Anima_preview3. This version was made beca...
SCOPE: Structured Decomposition and Conditional Skill Orchestration for Complex Image Generation
https://arxiv.org/abs/2605.08043
https://redd.it/1t9s8da
@rStableDiffusion
https://arxiv.org/abs/2605.08043
https://redd.it/1t9s8da
@rStableDiffusion
arXiv.org
SCOPE: Structured Decomposition and Conditional Skill...
While text-to-image models have made strong progress in visual fidelity, faithfully realizing complex visual intents remains challenging because many requirements must be tracked across grounding,...