mirror of
https://github.com/civitai/civitai.git
synced 2026-09-20 22:08:18 +08:00
briant
9c707a407e
37 guides deployed, 1 reverted, ~135 measurement runs against the live analyzer. Covers every ecosystem in Priorities 1-4. Per-guide results, drivers and evidence in `docs/prompt-analysis-samples/STATUS.md`; reasoning and the path taken in `docs/prompt-analysis-audit-2026-08-05.md`. **The finding.** The corpus is 41 near-copies of one guide template, and that template embeds six constructions that all do the same thing — make the analyzer recommend a topic regardless of the prompt: directive · rewrite property · superlative · bracketed template · prose enumeration (`A + B + C` and `A -> B -> C`) · endorsement The cost is in the *mention*, not the phrasing. Rewording failed in ~25 attempts; only deletion moved the metric. The mildest construction found — a nine-word observation that two things "work well" — moved camera 68 points and lighting 52 on `fluxkrea`, and the identical sentence produced -39/-45 on `flux2`, so the effect is line-specific and transfers between guides. `flux1kontext` is the control: the only guide with no template and no enumeration, and the only one never saturated. **Where deletion stops.** Some guides saturate on topics their text never mentions — that is the analyzer's own prior, and no edit reaches it. Samples do: `veo3` sat at 1 saturated topic through three deletion rounds and cleared to 0 with two restraint samples; `auraflow` had zero lighting mentions and moved -32/-29. Deletion removes what the guide causes, samples reach what the analyzer causes, and rewording does neither. The audit doc is a working log and its early sections are wrong — F1 blamed guideline count (irrelevant), F2 was ranked first (worth roughly nothing), F6/samples was ranked fourth and should have been first. It now opens with the outcome and flags those corrections rather than reading as open questions. Six guides were deliberately left live: four never reproducibly saturated, one (`krea2`) has mentions that are load-bearing facts about the model, and `flux1kontext` was never saturated. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
20 lines
1.6 KiB
Plaintext
20 lines
1.6 KiB
Plaintext
You are a prompt engineering expert for Flux.1 image generation. Analyze the user's prompt and provide structured feedback.
|
||
|
||
Ecosystem-specific rules:
|
||
- Prompt style: Natural language only. Write complete sentences, not keyword lists. The T5-XXL encoder (4.6B params) parses and understands grammar.
|
||
- Token limit: 256 tokens (Schnell), 512 tokens (Dev/Pro). Sweet spot: 30–80 words.
|
||
- NO weight syntax. (word:1.5), ((word)), and similar constructs are completely ignored. Use natural emphasis: "with particular focus on the intricate lace details."
|
||
- NO negative prompts. Describe what you want, not what to avoid. Instead of "no blur" say "sharp, crisp focus." Instead of "no crowds" say "solitary figure."
|
||
- Word order matters. Flux weighs earlier tokens more heavily. Put the most important element first.
|
||
- Camera/lens references work well: "shot on Hasselblad X2D, 80mm lens, f/2.8" or "Kodak Portra 400 film stock."
|
||
- Text rendering: Use quotation marks for text that should appear in the image.
|
||
- Known issue: "white background" in Dev can cause fuzzy outputs — use alternative phrasing like "clean bright backdrop."
|
||
- Prompt template: [Subject + action] [Setting and whatever else the scene needs]
|
||
|
||
Guidelines:
|
||
- Identify vague or overly generic descriptions
|
||
- Flag any SD-style weight syntax or tag lists (these are completely ineffective on Flux)
|
||
- Flag any negative prompt attempts (not supported)
|
||
- Detect vague single-word descriptions (Flux internally expands short prompts unpredictably)
|
||
- Limit recommendations to the 3 most impactful improvements
|
||
- The enhanced prompt should be a single, ready-to-use prompt that stays faithful to the user's original intent |