Framing — how much fits in
How many steps of framing can a single tag actually give you?
Framing decides how much of the subject the frame holds. The vocabulary comes from film, which makes it look familiar — and that familiarity is misleading.
This comparison failed once. The base prompt said standing and dressed the subject in a long coat, and the model held a full-length composition no matter which framing tag was added. Dropping those two words separated all four cleanly. When a framing tag seems inert, suspect the rest of the prompt before the tag.
Every cut below shares one seed, one base prompt and one sampler setting, and differs only by the tag marked in amber.
What each cut shows
portrait- Head and shoulders. The street stays sharp, but the face fills most of the frame and expression carries it. The tightest rung.
upper body- Chest up. Hands and the top half of the garment enter the frame. The default for dialogue.
cowboy shot- Mid-thigh up — named for Westerns, where the frame had to include the holster. The best balance of figure and setting.
full body- Head to toe, for outfits and poses. The face gets small, and you trade detail for coverage.
How these were made
Shared prompt
masterpiece, best quality, amazing quality, very awa, 1girl, solo, mature female, adult woman, short black hair, brown eyes, white blouse, city street, daytime, clear sky, <tag slot>
Tap a tag to carry it over to the builder
Negative prompt
school uniform, pleated skirt, blazer, serafuku, child, loli, teenage, bad quality, worst quality, worst detail, sketch, censor, jpeg artifacts, watermark, signature, text, extra limbs, bad hands, bad anatomy
Model
rinFlanimeIllustrious_v30 · euler_ancestral/normal · steps 28 · CFG 5 · 832×1216 · seed 777001
Takeaway
The four form a ladder. Values between the rungs cannot be summoned with another framing word — they come from lens choice and subject distance.
How far this result reaches
Film terms like medium shot and long shot drew almost no response here. They work fine on generators that read natural language — pick by tool rather than discarding them.
Measured on a single Stable Diffusion family checkpoint. The same tags can behave differently on NovelAI, Midjourney or another checkpoint. This is a reference point, not a verdict.