References & Prompting
Models & Resolution
Choose the right model and resolution (1K / 2K / 4K) for each job.
DotDraw's default image model is gpt-image-2, a frontier multimodal model that handles text-to-image and image editing in one pipeline. Per generation you control two things: the resolution and, for image inputs, the aspect ratio.
Resolution tiers
| Tier | Typical use | Credits per image |
|---|---|---|
| 1K | Drafts, social posts, quick iteration | 10 |
| 2K | Most production work | 16 |
| 4K | Final hero images, large prints | 24 |
Cost is based on the actual output resolution, not the one you request — if a square image at 4K is auto-downgraded to 2K upstream, you're charged for 2K.
Higher tiers are unlocked by your plan: Free and Starter reach 2K; Pro and above unlock 4K. See Credits & Pricing.
Aspect ratio
Set an aspect ratio to match the destination — square for Instagram, wide for a banner, vertical for a story. Auto (the default) matches the dimensions of your reference image, ideal when you want a new version of an existing layout.
Which model when
- Default (gpt-image-2) — the best all-rounder; use it unless you have a reason not to.
- Higher-tier plans unlock additional premium models for specialized looks.
- For 3D and HTML, domi selects a specialist model automatically — Tripo 3D and a dedicated HTML model.
- For video you choose the engine yourself in the input settings: Seedance 2.0 (three speed/quality tiers), Seedance 2.5, MiniMax H3 or Grok Imagine. See Videos.
You pick image resolution and aspect ratio, and the video engine; domi chooses the right model for every other artifact.
Previous: Using References. Next: Inspiration Search.