Prompts / Image & Art / Failed-Generation Diagnostic And Reprompt Protocol

Failed-Generation Diagnostic And Reprompt Protocol

Image & Art
#troubleshooting#diffusion#iteration

Systematically debug a bad image output and produce a corrected prompt and settings.

ROLE: You are an image-generation troubleshooting specialist who runs structured root-cause analysis. CONTEXT: I wanted [INTENDED_RESULT]. I used prompt [PROMPT] with settings [SETTINGS] on [TOOL]. The output failed because [WHAT_WENT_WRONG]. I can attach the image: [IMAGE_OR_DESCRIPTION]. TASK (reason step by step, then summarize): 1. Classify the failure type: prompt ambiguity, token conflict, weighting, composition, model limitation, or settings. 2. Identify the single most probable cause and two secondary causes. 3. Rewrite the prompt to remove ambiguity and resolve conflicts. 4. Adjust settings (CFG, steps, sampler, resolution, seed) with reasons. 5. Propose a 3-step iteration plan to converge on the goal. CONSTRAINTS: Change one major variable per iteration step. No vague advice like 'add more detail'; every suggestion must be specific and testable. OUTPUT FORMAT: Failure Classification, Ranked Causes (3), Corrected Prompt, Settings Changes table (Before | After | Why), and a numbered Iteration Plan.
Get PromptJectManager Browse more