Help With AI Prompting

[65sinking]

I've been trying for weeks to get Perchance to sink a character to the depth shown in the picture below, and the deepest I can get using a "submerged face deep in mud/quicksand" prompt, is about shoulder/neck deep. I was able do this a few months ago for some other projects, but now, I can't for the life of me get the character that deep. It Does anyone have some prompting advice/tips for accomplish this?

I know what you mean, I've also been struggling with getting Perchance to get people sinking to their chin or deeper. I usually use prompts like "face level sinking deep in quicksand" (with added brackets) to increase my chances of getting such results, but the output is inconsistent.

Strangely, I've found that adding in a prompt that describes the hair length or colour seems to interfere with the depth; if I keep generating images that show the subject only sinking to their shoulders, removing prompts like "long black hair" can result in the next round having subjects sinking deeper. Not sure why that's the case.

I'm sure you've also seen your fair share of failed attempts, such as the following:

Subject is sinking to their chin or even deeper, or have their entire face surrounded by mud, but they're embedded in a mound that rises above ground level:

photograph_ground-level_shot_b (100).jpeg

download (148).jpeg

Subject appears to be sinking to their chin, but only looks that way because there is a convenient pile of dirt that obscures part of their face:

photograph_ground-level_shot_b (89).jpeg

photograph_high-angle_close-up (7).jpeg

And of course, there are so many images generated that are technically sinking to that depth, but are unusable because the faces are horribly distorted.

[65sinking]

I've been getting some level of consistency with the depth by adding more prompts, like "upturned face", "head tilted upwards", "completely buried alive", "engulfed in mud", and "going under". I emphasise them by adding brackets (3 to 5 are ideal). Your mileage may vary though. Not sure if I'm overdoing it with the prompts, but you could try it out too.

Here are some examples:

photograph_high-angle_shot_1_5 (31).jpeg

photograph_high-angle_shot_bea (47).jpeg

photograph_overhead_shot_beaut (106).jpeg

photograph_high-angle_shot_bea.jpeg

overhead_photograph_beautiful_ (47).jpeg

[Theo]

I understand the frustration, I've experienced similar. Adding "looking up" to the framing part of the prompt, will often result in the head being tilted back. Then it's a case of getting the depth covered, I've had some success with having something along the lines of "sinks in quicksand until only the face is visible". Combining those two ideas may give a better chance of hitting what you want.

I haven't had a whole lot of time to keep working on the set in question as I've either been too busy with work or working on another major project I hope to post next month, but last night I did try a few of this ideas for a little bit. Still no joy

It's not really trial and error, it isn't that you are doing anything wrong. SD uses a seed number when generating images (you don't get to see this with Perchance), this is a large number (ten digits or so) and seems to be used to derive the noise pattern from which the resulting image is generated. Normally, a new seed is used for each generate, so you get a different result every time. This means that getting what you want is about pushing the image generation in the direction you want then generating until a good seed gets picked! The closer you can push the image generation prompt the more scope you have for getting a sympathetic seed.

Yes, a slot machine is probably a more accurate way to put it I guess. It's the luck of the draw, and my luck hasn't be super good lately (especially with these billion dollar lotto's over here—if I could just win one of them ) Oh, and BTW, I was going to mention, those were some nice pics you got, I particularly like the curly haired blonde on the beach And that first one you did was pretty cool too. I put that exact prompt into Perchance for the curly haired blonde—no where near as good as that one.

I know what you mean, I've also been struggling with getting Perchance to get people sinking to their chin or deeper. I usually use prompts like "face level sinking deep in quicksand" (with added brackets) to increase my chances of getting such results, but the output is inconsistent.

Strangely, I've found that adding in a prompt that describes the hair length or colour seems to interfere with the depth; if I keep generating images that show the subject only sinking to their shoulders, removing prompts like "long black hair" can result in the next round having subjects sinking deeper. Not sure why that's the case.

Yeah, "submerged face deep in (insert your desired substance: grey mud, quicksand etc)" is what I was using, but that's extremely hit and miss (like 99.99% misses). The problem with the long hair could be that the AI might think there's no room to depict it when the character is in too deep (even though, you're probably wanting that long hair spread out around her on the surface of the quicksand). I've had a similar problem if I leave "nude" or "naked" in a prompt for neck deep for example, there's not any naked body to see if it's under the mud, so the AI doesn't sink the character that deep.

Some great pics BTW, the first three on your second post are just about perfect