Drag across · scroll
0% read
The AI Image Soul Pipeline
FREEA Koda guide
Free guide: the art direction, prompt keywords and grain workflow that make AI images look shot instead of generated.
Everything you need, in one place
The generator
Midjourney
Your image engine. V8.2 is the current default model and it was tuned around aesthetics and Personalization, which is exactly the lever this guide pulls.
Create your accountThe motion
Kling AI
Kling Video 3.0 animates your stills up to 15 seconds and holds texture and elements steady across shots.
Create your accountThe alt motion
Dreamina (Seedance)
Seedance 2.5 for action shots, with long single-pass clips and a big pile of reference images.
Create your accountThe finish
DaVinci Resolve
Where you add halation, grain and gate weave so eight clips share one skin.
Download it- Time
- 90 minutes for a full 8-shot set
- Level
- Beginner, taste required
- Tools
- Midjourney, Kling AI, Seedance, DaVinci Resolve
- You end with
- A consistent set of frames that read as footage, not output
the diagnosis
Your images are not bad, they are undirected
Soulless AI has a signature. Once you can name it, you can delete it.
Look at the frames you posted last month. Subject dead center. Light coming from everywhere and nowhere. Skin with no pores. Zero grain, zero lens, zero weather. Every object equally sharp.
That is not a model problem. That is the model guessing, because you never told it where the light comes from, what camera is in the room, or what just happened.
The fix is art direction, not a better prompt generator. Four decisions do all the work: the film, the light source, the frame, and the texture that survives into motion.
Soul is specificity. Every vague word in your prompt gets replaced by the average of the internet.
midjourney + kling
Free preview
The rest of this guide is free.
Create a free account and keep reading right here. Takes 30 seconds, no card.
Create a free accountAlready have one? Log in