Tired of that yellow tint in ChatGPT images? Here’s how to fix it with simple color correction (www.techradar.com)

ChatGPT-generated images often show a subtle yellow/sepia cast that makes them look dated or less realistic — a quirk many users noticed is specific to ChatGPT’s image output. You can try to prevent it with prompt engineering (e.g., “neutral white balance,” “accurate skin tones,” “bright midday light” or negative prompts like “avoid yellowish/sepia tone”), but asking ChatGPT to “make this less yellow” usually recreates the image rather than simply color-correcting the original, producing a different result. A practical workaround is in-session post-processing: a short Python/OpenCV routine converts the image to LAB color space (L = lightness, A = green–red, B = blue–yellow), measures the B-channel’s positive shift from the neutral midpoint (128), and shifts that B channel back toward 128 (with clipping) to reduce yellow cast while preserving the original image content. This preserves composition and details instead of triggering a full re-render. The issue highlights how generator defaults and training biases can affect perceived realism and underlines the need for easy toolchain hooks (prompt controls or built-in color-correction) in image-generation pipelines — or switching to other models (e.g., Gemini’s Nano Banana) until the default bias is addressed.