Image & modality
Visual and multimodal craft — subject, light, camera, style, negatives, and grounded vision.
Soft lockYou can read this circuit. Practice in Studio requires Pro or Studio — Foundation stays free.
See Pricing →01
Visual attributesSubject, style, lighting, composition, quality, and negatives for image models.
Plate →Practice locked02
Subject-first framingLead with the primary subject and action before style flourishes so the model locks identity early.
Plate →Practice locked03
Lighting and atmosphereSpecify light direction, quality, and color temperature so mood is controllable instead of accidental.
Plate →Practice locked04
Composition and cameraCall shot type, angle, and framing so the canvas isn’t a random crop of the subject.
Plate →Practice locked05
Style anchorsPin 2–4 concrete style references (medium, era, artist-adjacent adjectives) instead of vague ‘cinematic’.
Plate →Practice locked06
Negative prompt disciplineA short Avoid list for known failure modes — not a dump of every artifact you’ve ever seen.
Plate →Practice locked07
Multimodal groundingWhen an image is input, force the model to cite visible evidence before inventing scene details.
Plate →Practice locked08
Reference image briefSeparate what to keep from a reference (identity, palette, layout) from what to change.
Plate →Practice locked