rodbishop on Nostr: For the basic art style, I might say "bold graphic art style illustration bright and ...
For the basic art style, I might say "bold graphic art style illustration bright and saturated ..."
Then for the central object and background and the colourway I might say "of an orange fantasy castle on a nondescript orange hill background..."
I'll try prompting composition "centred in/left of frame".
I might vary that several times until I have a good basic image to start working with.
Then I get into the editor and start masking off areas and regenerating them using seed images.
So for the workflow automation one I googled "elaborate machine with gears and screens" to find a suitable reference, and dropped that into the prompt with text "elaborate orange machine".
I find its very hard to get correct any secondary objects (frog with pick, octopus with sign, etc) or specific impressions (smiling eyes) in a single prompt, and so masking and regenerating the area is always better.
Hope that helps!
Published at
2025-01-07 20:54:54Event JSON
{
"id": "837c57c557254788a1a85c216c5f679c317f23770c17c3207c15510268268282",
"pubkey": "1bda7e1f7396bda2d1ef99033da8fd2dc362810790df9be62f591038bb97c4d9",
"created_at": 1736283294,
"kind": 1,
"tags": [
[
"e",
"06c844e9da18754d30e6656a3865341f7b6b13c6f94d19b14861be7c3048042c",
"",
"root"
],
[
"e",
"78a7bdd7f4981f728594249e5d909715ffd9f74dbe0305e1ed69790a3068c0b0"
],
[
"e",
"f29d0b6268bfaa288ba9b5fdfa1992de632e06265bdecc9efdff97b7ee1449b8",
"",
"reply"
],
[
"p",
"1bda7e1f7396bda2d1ef99033da8fd2dc362810790df9be62f591038bb97c4d9"
],
[
"p",
"bd4ae3e67e29964d494172261dc45395c89f6bd2e774642e366127171dfb81f5"
]
],
"content": "For the basic art style, I might say \"bold graphic art style illustration bright and saturated ...\" \n\nThen for the central object and background and the colourway I might say \"of an orange fantasy castle on a nondescript orange hill background...\"\n\nI'll try prompting composition \"centred in/left of frame\".\n\nI might vary that several times until I have a good basic image to start working with. \n\nThen I get into the editor and start masking off areas and regenerating them using seed images. \n\nSo for the workflow automation one I googled \"elaborate machine with gears and screens\" to find a suitable reference, and dropped that into the prompt with text \"elaborate orange machine\".\n\nI find its very hard to get correct any secondary objects (frog with pick, octopus with sign, etc) or specific impressions (smiling eyes) in a single prompt, and so masking and regenerating the area is always better. \n\nHope that helps!",
"sig": "d06067198244ebe68a923bc306995f40d9366d000bc534f5e80788325545475fe38dfaa9026821b18d8b3cc8a0f224762ca6a367b528b190912c145d576d4853"
}