ďťż
ďťż
1) First we need to generate referenced images as our start frames:
a) Agent settings:
Agent: Image generation
Text model: GPT-4o (but any text model will work well)
Image model: GPT Image 1 (this is the only model that will work this well for character referencing)
Attachment: upload a clear image of the person you want to reference
b) Agent prompt:
Convert my reference image of a person in XXX setting to A2 - maintain detail from reference - 16:9 aspect ratio
c) Style references
A2 - 80s/90s hi-NRG anime
A3 - 3D CGI kids movie animation
A4 - major studio cel-shaded kids movie animation
A5 - claymation
A6 - vector-style animation, style of corporate eLearning
2) Then we need to convert those images to animations
a) Video settings
Model: Veo 3 (This is the only model capable of audio/speech generation)
b) Video prompt
Animation using B2 as a start frame, depicting a person in XXX setting talking directly to camera in a XXX accent. In A2 style. In an XXX accent, they say to camera 'C2'
c) Script references
Include any text you want as a script but remember that each clip will be 8 seconds so each section should be concise. If you don't have much text here, you may have a long silence at the end of a clip and need to trim in post-production!
