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.
SenseTime previewed its next-gen multimodal model SenseNova-U1 Pro, claiming native 8K resolution output versus GPT-Image-2's 4K. The model uses a unified 'Understand-Generate-Action' architecture targeting professional design workflows. Direct comparisons showed U1 Pro outperforming GPT-Image-2 in an infographic, a scroll painting layout, a magazine spread, an academic poster, and a high-resolution storyboard. The model also generated the entire 20+ page shareholder meeting presentation end-to-end. Invite testing is slated to begin in July 2026.
A follow-up comparison tested Krea2 Turbo model with and without three different de-censor LoRAs. Two tiny LoRAs (about 200 bytes, changing only 2–3 weights) solely bypass the model's censorship filters without adding new styles or concepts. Results demonstrate that the built-in filters significantly nerf SFW image generation for facial expressions, bruises, body builds, emotions, and other natural features. The bypass LoRAs restore these lost capabilities by freeing concepts the model already possesses but underutilizes due to filtering. A trained SNOFS LoRA also removes filters but introduces additional styles, making it a less controlled comparison. The study uses fixed prompts and standardized parameters for fair evaluation.
The (Deno) Local LLM Loader is a new ComfyUI node that connects local LLMs (Ollama, LM Studio, llama.cpp, vLLM, or any OpenAI-compatible local server) directly into image generation workflows. It enables tasks like turning short ideas into detailed prompts, generating prompts from images via vision-capable models, and chaining multiple LLM steps (e.g., draft → review → final cleanup). The accompanying (Deno) Local LLM Reviewer node can pass or block image outputs based on review text, allowing iterative refinement without leaving ComfyUI. Designed as a local-first tool, it requires no remote API keys and keeps models loaded for efficient prompt chains. The nodes are available through ComfyUI Manager under "Deno Custom Nodes" and on GitHub.
A Reddit user reposted a workflow for creating 2x2 (4‑panel) cinematic storyboards that uses Krea2 Turbo for image generation and Gemma 4 12B for prompt engineering. The workflow includes a node for LM Studio with a carefully crafted system prompt that makes Gemma produce highly detailed Krea2 prompts. Krea2 Turbo currently struggles with larger grids and often generates asymmetric panels, so the user is developing a custom node to fix this, but it is not yet included. Only a simple scene description is needed to generate a storyboard; examples such as a children’s fantasy film and a hacker scene are provided. The workflow file is available via Google Drive.
A Reddit user reported spending three hours with Krea 2 Turbo and found it extremely impressive. All images were generated in WanGP at 1440p resolution. No masking, layering, or LoRA techniques were applied. The user shared a link to source images and prompts but provided no further technical details.