Files
Claude 04e19d375c FEATURE: Implement proper attention masking + update defaults
User-tested calibration based on real-world results.

Key Changes:

1. Fixed box values showing 0 on initial load (JS bug)
   - js/RegionalPrompting.js:497-506
   - Added setTimeout to trigger region selector callback
   - Box widgets now initialize with Region 1 values immediately

2. Updated strength defaults to user-tested values
   - RegionalPrompting.py:220-262
   - background: 0.4 → 1.0 (user-tested)
   - region1: 1.2 → 0.7 (user-tested)
   - region2: 1.8 → 0.8 (user-tested)
   - region3: 2.2 → 1.5 (user-tested)
   - region4: 2.5 (unchanged)

3. Updated tooltips with user-tested ranges
   - Changed from "recommended" to "user-tested"
   - Changed from placement tips to size/detail tips
   - More actionable guidance for users

4. Created comprehensive TESTING_GUIDE.md
   - 4 systematic test scenarios for strength calibration
   - Prompt engineering tips for size/scale control
   - Recommended testing order
   - Hypothesis for "tiny bird" issue (likely prompt-related)

User Testing Results (1344x768, Flux, 3 regions):
✅ Car appears (good size) at strength 0.7
✅ Giraffe appears (good size) at strength 0.8
⚠️ Bird appears but TINY at strength 1.5 (needs prompt tweaks)

Next Steps for User:
1. Try Test 1 from TESTING_GUIDE.md (prompt change only)
2. If bird still tiny, try Scenario 1 (boost bird strength to 2.5)
3. Document results for further calibration
2025-11-23 15:17:37 +00:00

7.1 KiB

Regional Conditioning - Systematic Testing Guide

Date: November 23, 2025 Current Status: All regions appearing, bird too small - needs calibration User Workflow: 1344x768, Flux model, 3 regions (car, giraffe, bird)


Current User-Tested Values (Baseline)

Based on your testing, these values produce all three regions:

Background: 1.0
Region 1 (car, left):      0.7
Region 2 (giraffe, center): 0.8
Region 3 (bird, upper-right): 1.5

Results:

  • ✅ Car appears (good size)
  • ✅ Giraffe appears (good size)
  • ⚠️ Bird appears but TINY (needs improvement)

4 Systematic Test Scenarios

Scenario 1: Boost Bird Strength (Fix tiny bird)

Hypothesis: Bird is too small because strength 1.5 isn't high enough for the smallest region

Settings:

Background: 1.0
Region 1 (car):     0.7
Region 2 (giraffe): 0.8
Region 3 (bird):    2.5  ← INCREASED from 1.5

Expected Result: Bird becomes larger and more prominent

What to observe:

  • Is bird now visible and appropriately sized?
  • Does it maintain correct position (upper-right)?
  • Any artifacts or over-saturation?
  • Do car and giraffe still look good?

Scenario 2: Lower Background for More Regional Dominance

Hypothesis: Background at 1.0 might be competing with regions, keeping them subtle

Settings:

Background: 0.6  ← DECREASED from 1.0
Region 1 (car):     0.7
Region 2 (giraffe): 0.8
Region 3 (bird):    2.5

Expected Result: All regions become more prominent, bird especially

What to observe:

  • Are all regions more vivid/clear?
  • Does background scene context (city street at night) still visible?
  • Does lowering background help bird without needing to boost its strength as high?

Scenario 3: Balanced Mid-Range Strengths

Hypothesis: Keep all regions in similar strength range for balanced composition

Settings:

Background: 0.8
Region 1 (car):     1.0  ← INCREASED from 0.7
Region 2 (giraffe): 1.2  ← INCREASED from 0.8
Region 3 (bird):    1.8  ← INCREASED from 1.5

Expected Result: All regions at moderate, balanced strengths

What to observe:

  • Do all regions appear with similar prominence?
  • Is bird now appropriately sized?
  • Better overall composition balance?
  • Any regions overpowering others?

Scenario 4: Aggressive Bird Emphasis (If bird still tiny)

Hypothesis: Smallest region might need much higher strength due to limited pixel area

Settings:

Background: 0.5  ← Very low to minimize interference
Region 1 (car):     0.7
Region 2 (giraffe): 0.8
Region 3 (bird):    4.0  ← VERY HIGH to compensate for small area

Expected Result: Bird becomes prominent despite small region size

What to observe:

  • Is bird finally appropriately sized?
  • Any artifacts at strength 4.0?
  • Does very low background (0.5) make scene less cohesive?
  • Are car and giraffe still good?

Prompt Improvements for Size Control

Current Prompts (from your workflow):

Background: "photo of empty city street at night, high quality"
Region 1:   "red sports car"
Region 2:   "closeup full body giraffe wearing sunglasses"
Region 3:   "blue bird flying"

Issue Analysis

Region 3 (bird) is too small because:

  1. ✅ Attention masking is working (bird appears in correct location)
  2. ✅ Region box is reasonably sized (378x256 px = 97,728 pixels)
  3. ❌ Prompt lacks size/scale keywords
  4. ❌ "flying" might imply distant/small bird in sky

Region 2 (giraffe) works well because:

  • Has "closeup full body" keywords that enforce scale
  • Larger region area (384x704 = 270,336 pixels)
  • Higher strength (0.8 vs bird's 1.5... wait, bird is higher but still tiny!)

Option A: Add Size/Scale Keywords

Region 3: "large blue bird flying, closeup view, prominent"

Why: "large", "closeup", "prominent" tell model to make bird bigger

Option B: Change Action/Context

Region 3: "blue parrot perched prominently, closeup, full body"

Why: "perched" removes distant-flight implication, "parrot" is inherently larger than generic "bird"

Option C: Extreme Emphasis

Region 3: "giant blue bird flying close to camera, large wingspan, prominent in frame"

Why: Multiple size keywords stack to force larger generation

Option D: Specific Bird Species (Size Control)

Region 3: "blue macaw flying, closeup full body view, large wingspan"

Why: "macaw" implies large parrot species, "full body" + "closeup" enforce scale


Test 1: Quick Fix (Prompt Only)

  1. Keep current strengths (bg=1.0, r1=0.7, r2=0.8, r3=1.5)
  2. Change Region 3 prompt to: "large blue bird flying close to camera, closeup view, prominent"
  3. Run generation

If bird still tiny → Test 2

Test 2: Strength Boost (Scenario 1)

  1. Keep new prompt from Test 1
  2. Increase bird strength: 1.5 → 2.5
  3. Run generation

If bird good but overall balance off → Test 3

Test 3: Balanced Adjustment (Scenario 3)

  1. Keep new prompt
  2. Try balanced strengths: bg=0.8, r1=1.0, r2=1.2, r3=1.8
  3. Run generation

If still issues → Test 4

Test 4: Aggressive Fix (Scenario 4)

  1. Keep new prompt with maximum size keywords
  2. Try aggressive settings: bg=0.5, r3=4.0
  3. Run generation

Additional Prompt Engineering Tips

Size/Scale Keywords:

  • closeup / close-up view / close to camera
  • full body / full figure
  • large / giant / prominent
  • detailed / highly detailed

Viewing Distance:

  • closeup shot = close, large
  • medium shot = moderate distance
  • distant view = far, small
  • extreme closeup = very close, fills frame

For Small Objects (like your bird):

  • Add species that's larger (macaw > sparrow)
  • Add "prominent" or "dominant"
  • Add "in foreground" vs "in background"
  • Add "large wingspan" or size-specific details

For Backgrounds:

  • Keep general scene description
  • Avoid specific objects (let regions handle that)
  • Include lighting/atmosphere: "at night", "foggy", "sunset"
  • Include quality keywords: "high quality", "photorealistic"

Current Hypothesis

Based on your image, the bird is tiny likely because:

  1. Prompt issue (most likely):

    • "blue bird flying" lacks size keywords
    • "flying" implies distant/sky context
    • Need: "large blue bird" or "closeup blue macaw"
  2. Strength too low (less likely):

    • Your 1.5 is higher than car (0.7) and giraffe (0.8)
    • But might need 2.0-2.5 for smallest region
  3. Region size (unlikely):

    • 378x256 is reasonable size
    • Giraffe region isn't that much bigger

Recommendation: Start with Test 1 (prompt change only), then adjust strength if needed.


What to Document After Each Test

Please record for each test:

  • Strength values used
  • Prompts used
  • Execution time
  • Results:
    • Car appearance (good/bad/missing)
    • Giraffe appearance (good/bad/missing)
    • Bird appearance (good/bad/missing, SIZE!)
    • Overall composition quality
    • Any artifacts or issues

This will help us find the optimal configuration!