Handmade collaged images – whether analog, digital or hybrid – can be used as a guide to create consistent compositions in photorealisitc or any illustrative or surreal style. The handmade composition is a basic input for a proper image2image workflow using one of the bigger diffusion models to create a visual consistent image with fine details. With variation in denoise value, proper prompting and manual visual input, we can guide the diffusion process into very custom articit and specific ways.
FLUX_KREA_TURBO WORKFLOW FOR COMFYUI: FLUX_KREA_MZ
simple image 2 video interpolations
With proper startframe image design, we can use video diffusion models to deliver plausible and interesting video interpolations. Both visual representation like motion blur or plausible constellation of elements in combination with a proper prompt for the video generation is key here!
frontal portrait of person with sculpture on the head . the sculpture is from moving car parts. tires are spinning, parts slightly move. head turns a bit. static handshake camera.
advanced guided interpolations
Some video diffusion models come with the feature of startframe and endframe setup. This means, we can define a detailed choreography of what should happen in the image. Keep in mind, that the changes between startframe and endframe should be similar and comparable – otherwise the generative model starts to halucinate and loose track. ( this migh be an interessting effect too 🙂 )
portrait of bald male person with excentric sculpture on the head. wings move, hamsters run, gloves and fingers move, mouth opens and shows black teeth. handheld static camera.