Let's go over how I generated the images in my manga from the framework that I introduced. Now that we have gone over the simple Framework, here is again my Framework for generating images:
1. Image Prompt (Paste the image link to reference here), you may post several
2. Composition (aerial view, closeup, birds eye, side angle, etc)
3. Medium (‘photo/image/illustration of a XYZ,’ you can also add style adjective before medium like ‘renaissance style painting of a…’)
3. Subject (grey haired old man with sweatshirt etc)
4. Context (jumping around / inside a)
5. Environment (café / forest), can add adjective.
6. Lighting or Colour (these can be embellished in 2 or 3 or at the end)
7. Image weighting / Text Weightings / Style Weightings
8. Aspect Ratio
9. Engine pick
Here an example would be:
/image https://s.mj.run/wxHseMJzCOs https://s.mj.run/aTgJE_gN0SM hungry cyborg with red face shield eating a bowl of ramen noodles in a small noodle shop image by Enki Bilal::1 --iw 2.0 --ar 16:9 --niji 5
These 2 were my reference images, the left being an illustration of a cyborg and the right being of an anime character eating ramen at a ramen shop. The command @Enki Bilal::1 means I want the image to be in the style of this particular artist with a weighting of 1, --iw 2.0 means I want the reference images to carry double the weighting and --ar 16:9 is the command for the aspect ratio with the last –niji 5 being the command for the anime engine that I want to use. The 2 reference images:
This was the result:
Since the character had a hollow face which couldn't be used but more or less had the Cyborg look that I needed, I then used the above image just generated as the new 'reference image' and together with my original cyborg image, re-engineered the image combining these two:
This was the command:
/imagine https://s.mj.run/wxHseMJzCOshttps://s.mj.run/UA_UphAhe08 hungry cyborg with red face shield eating a bowl of ramen noodles in a small noodle shop image by Enki Bilal::1 --iw 2.0 --ar 16:9 --niji 5
And the final result used in the manga:
It is with the same framework and method that I could get images like this:
And this:
