In continued pursuit of Cassandra, the Mythical Prophet of Troy

in #hive-1224722 years ago

image.png

I promise that I will eventually be done with this subject matter, but for the time being, it is one that continues to challenge me. The above image is about where I left off last time, but not exactly. It is a crude representation of the vision that exists in my mind's eye.

That vision is of a incredibly detailed, vividly illustrated woman, the prophet, Cassandra, standing before (or among) the burning ruins of the city of Troy.

Her prophecies were not believed as part of the curse placed upon her, so she should have a mixture of despair and sadness in her pose or expression, and ideally, there will be some figures scurrying around in the background.

Perhaps a dramatic break in the clouds or the smoke, from which the gods cursed her.

It is an ambitious vision to do such a thing with nothing but AI and a little guidance, but I will eventually succeed to represent this. My vision may morph and alter along the way, but here I will resume the quest to find my artistic gaze that I started with the previous couple of posts on this subject matter.

With the refined prompt from the last set of iterations, I got the following as starting images.

image.png

The first three are absolutely garbage options that no sane or rational artist would consider investigating further, but the fourth one shows some legitimate promise in terms of eyes, the background, and the composition, if basic.

image.png

An enlarged version appears to be a fairly roughly painted version of the thumbnail, and that's not a terrible thing, what happens when I try to tidy it up further by using a different sampling methodology?

image.png

The lighting here is a little too "high key" for my liking, I would prefer it to be much darker, with more smoke. The other issue is that the woman in the foreground, our elusive Cassandra, appears to be simply stamped into the scene, she doesn't really seem like she's "part" of the scene, but rather just composited on top of it.

Furthermore, the characters in the background appear to... have vanished in the upscaling and resampling process which is slightly disappointing. I will try the same prompt again, and see what a few different random seeds can get me.

image.png

image.png

image.png

The next set of thumbnails are all pretty similar in terms of compositional cues. There's also some really strange and bizarre facial structures.

There's a problem with the next image, it has no face at all:

image.png

image.png

I think I need to change my prompt to get some more coherent renderings of the image I am after. Perhaps I am using too many words.

image.png

I adjusted the prompt, and got some slightly different results. They still need many more random seeds to be run in order to see what magic is hidden in the world of random noise, but its a task I stick to, because the Art is out there, hidden somewhere.

image.png

image.png

image.png

Can you tell where I swapped from Midjourney to Stable Diffusion, using the same prompt? I'll let you guess. I'm rather impressed with the outcome of the hands. Much better than Mid Journey.

There's so much detail in the faces, but the backgrounds are now lacking, so I will need to modify the prompt a little bit more.

A highly detailed cinematic headshot portrait color photograph of a beautiful ancient greek woman, shot in the burning hallway of a ruined temple with figures running around in the background, ultra realistic, depth, beautiful lighting, by richard avedon and annie leibovitz and arnold newman, photorealistic, hyperrealistic, octane, epic composition, hasselblad camera, 35 mm, sharp focus, perfect facial symmetry

image.png

What a strange arm. What incredible texture.

image.png

These are certainly tortured characters, but the background. The backgrounds need much, much more work.

image.png

Such impressive textures on display, and amazing lighting on show, too. I might be more literal with the prompt from this point forward, to ensure the point is understood.

she is a beautiful woman standing in front of the burning city of troy, cowering in the hallway of a crumbling temple, A highly detailed cinematic headshot portrait ultra realistic, depth, beautiful lighting, by richard avedon and annie leibovitz and arnold newman, photorealistic, hyperrealistic, octane, epic composition, hasselblad camera, 35 mm, sharp focus, perfect facial symmetry

image.png

I'm still getting amazing results and interesting characters, with absurd details and textures, but the backgrounds are not doing themselves any favours. The best result from this batch was probably this image:

image.png

The issue is this: it isn't in colour, and the gown is not of the right style. Maybe if I use the same seed with a further modified prompt.

she is a beautiful woman standing in front of the burning city of troy, cowering in the hallway of a crumbling temple, A highly detailed colour cinematic headshot portrait dramatic lighting, smoke in the air, ultra realistic, depth, beautiful lighting, by richard avedon and annie leibovitz and arnold newman, photorealistic, hyperrealistic, octane, epic composition, hasselblad camera, 35 mm, sharp focus, perfect facial symmetry

I forgot to make modifications about the dress, but I did ask for colour this time around, so what will it be?

image.png

colour image of a beautiful ancient greek woman standing in front of the burning city of troy, cowering in the hallway of a crumbling temple, A highly detailed colour cinematic headshot portrait dramatic lighting, smoke in the air, ultra realistic, depth, beautiful lighting, by richard avedon and annie leibovitz and arnold newman, photorealistic, hyperrealistic, octane, epic composition, hasselblad camera, 35 mm, sharp focus, perfect facial symmetry

Colour. Colour. Colour. Maybe I need to say Color? I'll try that next. Updated the information about the outfit. By defining what sort of woman. Got a different result entirely...

image.png

I'm still using the exact same seed as the other recent images. This means that they should be substantially similar, but they're a bit different - which is fine, but it is still not the image I want.

detailed color photograph of a beautiful ancient greek goddess standing in front of the burning city of troy, cowering in the hallway of a crumbling temple, A highly detailed color cinematic headshot portrait dramatic lighting, smoke in the air, ultra realistic, depth, beautiful lighting, by richard avedon and annie leibovitz and arnold newman, photorealistic, hyperrealistic, octane, epic composition, hasselblad camera, 35 mm, sharp focus, perfect facial symmetry

image.png

I guess goddess instead of woman means a statue. :D

a beautiful ancient greek woman looking away from the burning city of troy in the hallway of a crumbling temple, A highly detailed color cinematic headshot portrait dramatic lighting, smoke in the air, ultra realistic, depth, beautiful lighting, by richard avedon and annie leibovitz and arnold newman, photorealistic, hyperrealistic, octane, epic composition, hasselblad camera, 35 mm, sharp focus, perfect facial symmetry

Again, the same seed: 8921383

image.png

Time to try some different seeds... plus, we lost the colour again.

image.png

Certainly a rendition of a goddess here, but the details are not as sharp as some of the prior seeds. Some are good, some are bad, it seems. Will have to keep iterating. This could be a long night of pressing the Make Image button in Stable Diffusion UI. :D

Changing the prompt again, after about another 50 iterations I was not happy with:

highly detailed cinematic portrait color photograph of a beautiful ancient greek woman perfect facial symmetry, flames behind, shot in the burning hallway of a ruined temple on fire in flames, figures running around in the background, ultra realistic, depth, beautiful lighting, by richard avedon and annie leibovitz and arnold newman, photorealistic, hyperrealistic, octane, epic composition, hasselblad camera, 35 mm, sharp focus

image.png

Immediately, that's working more to the intention... never mind the character. I think this prompt is more of a winner. Big shame about the coherence of the characters on display, though.

image.png

This next image is a very interesting take, with a completely different colour palette.

I do not mind the way the figures in the background bleed into the image, they're almost... not there. What a wonderful rendering!

image.png

Then there's this:

image.png

Going to generate another batch of 15, then call it a night. Will see where I end up, and what my next starting point will be in terms of a prompt to work on. This is tougher than I thought, but it is getting very late here :D

What an incredible expression.

The outputs from this newer, slightly modified prompt are now suddenly amazing.

image.png

7440697

image.png

7440705

I need sleep. More of this tomorrow.


Want more content from me?

Witness my futile efforts to play my Steam Game collection in alphabetical order.

Are you aware that I love photography? Check out my work in a collection.


If you haven't started playing Splinterlands, you should do that immediately. It's very good fun.

If you want to see my Splinterlands antics and rants live, Find me on Twitch

If you prefer sleeping in your designated time zone, go watch replays on YouTube.


Thanks as always for your time!