6Prompt vocabulary — what the chat translates for you
Type the left side. The chat looks at the image, derives the right side, and queues the render. You don't have to learn the lexicon — the translation is the product. The right column is shown so you can verify the model picked the right deltas, not so you have to type them.
"Warm the shoulders"
color_grade.highlights.r_shift +6 to +10 (capped)
"Smaller face / softer jaw"
liquify_face.jaw_taper bounded by face-bbox caps
"Make the skin cleaner"
freq_sep_radius 9 + neural_filter_skin_smooth 0.3
"More punchy"
camera_raw.contrast_pct +6 to +12, clarity_pct +8 to +14
"Lift the shadows"
camera_raw.shadow_lift +15 to +25
"Recover the highlights"
camera_raw.highlight_recovery -10 to -25
"Cooler / less yellow"
camera_raw.wb_temp_shift -1.5 to -3 (Kelvins)
"Golden hour glow"
split_tone highlight_color (248, 205, 158), intensity 0.18-0.25
"Make it natural"
inversion pass — pull every knob 30-50% back, re-evaluate
"Less aggressive"
output_opacity 0.65-0.80 (final layer blend)
"Eyes look tired"
generative_fill_under_eye true (ai_assists block)
"Editorial / magazine feel"
contrast +8, vibrance +6, split_tone shadows toward (56,74,93)
"Make me slimmer / improve body"
liquify_body.waist_taper -8 to -14px, garment fabric compression smoothing, shoulder-to-waist transition cleanup (personal-portrait-natural lane caps body deltas)
"Improve face / more attractive"
under_eye_lift 5–10%, eye_open +3–5%, jaw_transition_softening, +0.3–0.5EV face fill, clarity +6–10 on eyes
"Subtle edits / clean it up"
+0.3EV global lift, clarity +8 on center of interest, -0.15 saturation red channel, minor split-tone warm highlights