Assemble an image prompt
Set up your cast, pick one master prompt, tick the modules, press Assemble. Everything runs on this page. No photo and nothing you type ever leaves the device.
Packs
Load a folder of JSON pack files to add or replace anything below. Loaded packs are kept on this device, so you only need to load again after you edit a file. You can also drag files onto this page.
Cast
How many characters recur across your images. Photos are only ever shown here, as a reminder of which slot is which. Attach the real photos to your message in the AI itself.
Master prompt
pick oneThese are not compatible with each other. Each one tells the model what kind of object it is making.
Modules
0 selectedStack as many as you like. They plug into the master without changing it. Modules that do not suit your chosen master are hidden.
Dials
These do not print numbers into the prompt. Each position emits different wording.
Saved setups
Keep a combination you liked: the master, the ticked modules, the dials and the text boxes. Setups are stored on this device only, and never include your cast or photos. Modules are remembered by their wording rather than their number, so a setup still loads correctly after you add or remove packs.
Generate one image at a time. Reject anything with a spelling error rather than accepting it, since a rerun usually fixes it. Rejected images stay in the conversation and influence the next one, so when a detail sticks, edit your earlier message to branch rather than replying again. After roughly five images in one chat, start a fresh one and re-attach the reference photos.