Photoroom engineers found that classifier-free guidance scales above 7.5 caused oversaturation and clipped highlights in their SDXL product renders. They fixed the issue by adding CFG rescale at 0.7 a…
Researchers introduced block-sparse featurizers (BSFs) to capture visual concept manifolds in neural network representations, showing that concepts are low-dimensional geometric structures rather than…
Runflow benchmarked 8 image generation models across 12,000 real inference jobs, scoring each on quality, cost, and latency. Flux.2 [dev] achieved a composite quality score of 95, matching or exceedin…
ComfyUI's built-in HTTP server on port 8188 accepts workflow JSON in API format, enabling image generation automation without the GUI. Developers can queue prompts, poll for completion, and download r…
A new benchmark dataset, the Social Media Robustness Benchmark, evaluates how deepfake detectors perform on images re-encoded by social media platforms like Instagram, Facebook, TikTok, and X. The dat…
Photoroom reduced its vision-language model inference costs by approximately 62% within three weeks by deploying Bifrost as a semantic caching layer in front of the VLM step of its product-photo diffu…
Photoroom's SDXL LoRA fine-tuning for a product photography model silently corrupted its adapter weights over six days due to a bf16 gradient accumulation issue. The custom training loop, forked from …
Researchers have developed Diff-Instruct with Diffused Reward (DIDR), a data-free trajectory-level alignment framework that propagates reward-tilted clean-image distributions across all noise levels t…
The "CUDA out of memory" error in Stable Diffusion WebUI is often caused by configuration issues rather than insufficient GPU hardware, particularly due to PyTorch's memory allocator failing to releas…
Slow single-image diffusion model inference is primarily caused by kernel launch overhead and attention memory traffic, not by a lack of computational power. It recommends using `torch.compile` with `…