Open tanghengjian opened 2 months ago
I don't understand, sorry
when draw the first person's face image, should the background be taken care in first person's mask area? when draw the second person's face image, the background also be taken care in second person's mask area? in a word, how the background of the junction of two different mask area looks so smooth, because the painted area will be abrupt if they are painted in different latent space?
please take a look @cubiq
the model will try to merge the two automatically
so you mean the background really painted in left and right area, and merged finally? Which model is responsible for merging two part's image? in latent data or real image data? have confuse for a lot of days. Sorry for my questions...
In "Multi-ID" workrflow, how the background image keep consistent in different mask area? Basic workflow: 1、Apply InstantID, person 1 process, 2、Apply InstantID, person 2 process, 3、ksampler