A user built an AI-assisted animation pipeline using Blender and ComfyUI, employing LTX 2.3 as an alternative render engine. They modified the official IC-LoRA workflow to include both first and last frame conditioning, and used a custom control video combining depth and AO passes from Blender. With only the distilled LoRA, motion and composition were stable but textures faded away from guide frames; adding IC-LoRA preserved textures but caused composition drift, character misplacement, and overall instability. The user seeks a technical understanding of how IC-LoRA interacts with the distilled model and guide frames to explain this trade-off.
A user discovered a method to generate an effectively unlimited number of high-resolution images with consistent characters by having the language model reconstruct the full semantic state of every frame from scratch, rather than relying on image-based memory like IPAdapter or character LoRAs. The workflow involves writing a single story prompt containing detailed character sheets and scene descriptions; a Qwen VLM node splits the story and rewrites each character’s description completely for every panel before feeding it to Krea 2. The approach yields surprising consistency, requiring no reference images or multi-panel tricks. The method works with Krea 2 and likely other capable models, and the full ComfyUI workflow is publicly shared for others to try with Flux, HiDream, or Seedream.
The Phonediffusion iOS app enables fully on-device Stable Diffusion workflows with a ComfyUI-like interface. Users can chain multiple steps: txt2img, img2img, upscaling, and background removal. The app runs locally and offline (airplane mode) and performs fast on an iPhone 17, while still usable on a 6-year-old iPhone 12 at approximately half the speed. The developer shared a demo of product photo cleanups and invites suggestions for additional workflows. The app is available on the App Store.
A Reddit user shared their experience of old image-generation prompts from 2022 now producing the intended results with KREA 2. The generation reportedly completed in a single pass taking 15 seconds. The post includes links to the prompt and workflow files as well as a LoRA on Civitai.
A Reddit user introduced a technique to apply Krea 2's safety filter LoRA only on select parts of the image generation process, aiming to reduce the visible quality degradation caused by the filters. Inspired by a prior extraction of the filter's internal values, this approach shows some improvement in output but slows down generation. The optimal implementation and overall effectiveness remain uncertain.