Flux 2 Max Prompting Guide

Flux 2 Max is Flux's text-to-image model on Versely. This page is its structured prompting reference: the 3 parameters its schema actually exposes, the t2i technique that applies to it, copy-ready templates.

Everything here is grounded in the same sources Versely's agent reads — the model's input schema, the flux family rule its prompt enhancer applies. Where a line is general craft advice rather than a documented fact about Flux 2 Max, the page says so.

What Flux 2 Max wants

The exact input surface, from the same schema the Versely agent fetches with get_model_input_schema before every generation.

The rule Versely's enhancer applies

Versely's prompt enhancer carries a per-family rule for flux models, applied automatically whenever it rewrites a prompt for Flux 2 Max. Verbatim:

“Flux models excel with natural language descriptions. Be specific about composition, color palette, and mood. Flux handles photorealistic prompts very well — include camera details like lens type, aperture, and film stock for photographic styles.”

Technique that applies here

Text-to-image: composition, lighting, style; family tips (Midjourney/Flux/Imagen/etc.) drive syntax

• The enhancer also injects a family-specific MODEL_TIPS block once it recognizes the model name (substring match, first hit wins, checked in this order: midjourney, flux, dall, imagen, stable, kling, veo, minimax, wan, seedance, sora, hunyuan). On this family that means: Midjourney V7/Niji 6 get 'comma-separated descriptors... artistic references, lighting descriptions... and medium descriptors... avoid generic filler words'; Flux (2 Max, Flex, Klein 9B, Flash, 1.1 Pro, Schnell) gets 'natural language descriptions... specific about composition, color palette, and mood... include camera details like lens type, aperture, and film stock for photographic styles'; Imagen 4 gets 'detailed natural language prompts... subject description, environment, lighting conditions, and artistic style... explicit about what you want in the scene'. Kling Image 3.0 and Wan 2.6 Text to Image match their family's tip too, even though it's written primarily for the video siblings — Kling's says 'clear scene descriptions including camera movement (pan, zoom, tracking shot), subject action, and environment details. Specify motion direction and pacing.' A still image has no motion to specify, so use the 'clear scene description... environment details' half and drop the camera-movement language; it has nothing to act on here.
• Pick aspect ratio / size from the model's actual enum rather than describing it in prose. Cosmos 3 Super and Recraft 4.1 both take the same six image_size presets (square_hd, square, portrait_4_3, portrait_16_9, landscape_4_3, landscape_16_9); Flux 2 Max's resolved schema takes that same six plus a 7th value, auto; Seedream 4.0 uses the six-preset list too, plus a custom {width, height} object (960-4096px, ~2048x2048 default); Grok Imagine Image Quality and Krea 2 Medium instead take an aspect_ratio string enum (Krea's list includes a 2.35:1 cinematic option Grok's doesn't have).
• Several of these pages are also the home of an edit-capable gate variant, not pure generation only: Flux 2 Max's resolved schema is actually the Flux 2 Pro Edit endpoint (image_urls + prompt + image_size) because it folds in Flux 2 Max Edit / Flux 2 Pro Edit; Nano Banana folds in Nano Banana Edit and Nano Banana Pro Edit; GPT Image 2 Text to Image folds in GPT Image 2 Image to Image; Wan 2.6 Text to Image folds in Wan 2.6 Image to Image. Attach a reference image on one of these and you're editing that image, not generating fresh — know which mode a given call is in.

Copy-ready templates

Replace the bracketed slots; each template says when it's the right shape.

[SUBJECT], [ACTION OR POSE], in [SETTING/ENVIRONMENT], [LIGHTING] lighting, [COLOR PALETTE], [STYLE OR MEDIUM], [MOOD/ATMOSPHERE]. (Pick aspect ratio/size from the model's own preset field — the schema takes an enum value, not a ratio typed into the prompt.)

[SUBJECT], [SETTING], shot on [FILM STOCK OR CAMERA TYPE, e.g. 35mm film / medium format digital], [LENS, e.g. 85mm portrait lens], [APERTURE, e.g. f/1.8 shallow depth of field], [LIGHTING], [COLOR PALETTE]

Mistakes that waste generations

• Padding every prompt with quality-booster tag lists ("masterpiece, best quality, highly detailed, 8k"). That convention belongs to the 'stable' MODEL_TIPS family — comma-separated tags plus quality boosters — and nothing in this family resolves to it; on Flux, Imagen, Midjourney, or any of the fallback-tip models it just reads as filler the model has to parse around.
• Free-typing an aspect ratio or size string outside the model's real enum — e.g. asking for 'a wide 2.5:1 banner' on a model whose aspect_ratio list tops out at 21:9, or writing 'widescreen' where the field wants the literal preset value landscape_16_9. The schema enum is exact; the model isn't interpreting loose ratio language.
• Writing an edit instruction ("make the sky sunset-orange") into one of the plain-generation pages that also covers an edit-capable gate variant (Flux 2 Max/Pro Edit, Nano Banana Edit, GPT Image 2 Image to Image, Wan 2.6 Image to Image) without attaching a reference image. With no image_urls supplied, you get a brand-new unrelated image, not an edit of anything.

This guide also covers

These siblings share Flux 2 Max's prompting-relevant input surface, so their prompting URLs resolve here — tier and pricing differences live on their own model pages:

Frequently asked questions

Related prompting guides