JD's bazaar

[doe.1971]

Spectacular image. Can you do close ups of each of the three women?

They will become the 'whipped-women-on-the-post' typical images, the kind you find online plenty, most at better quality than what I can do ... Cui bono?

[bleumune]

Spectacular image. Can you do close ups of each of the three women?

They will become the 'whipped-women-on-the-post' typical images, the kind you find online plenty, most at better quality than what I can do ... Cui bono?

What?!? Better than you can do?! Ridiculous!!!

If that was a true statement, then why are your posts on my check-out list whenever I find them? You're correct about finding many posts of similar topics, but they are very seldom "better quality" (unless they're made by Jucundus, MadBob, and a very few others).

Better work on that inferiority complex, bro! Get to work on those close-ups!!

[doe.1971]

What?!? Better than you can do?! Ridiculous!!!

If that was a true statement, then why are your posts on my check-out list whenever I find them?

Because you're a perv, that's why!

Get to work on those close-ups!!

I tried, trust me. Unremarkable results. Textures are rough, the bad bendings are showing, shadows are too harsh, light looks fake. I've tried, nothing worth salvaging there. But I love the appreciation, that's sure!

[xs70]

Get to work on those close-ups!!

I tried, trust me. Unremarkable results. Textures are rough, the bad bendings are showing, shadows are too harsh, light looks fake. I've tried, nothing worth salvaging there. But I love the appreciation, that's sure!

I don't have any 3D skills, but if I did, I would use something like Daz 3D to create my scene, then pass the image to a Stable Diffusion model for the rest (texturing, shadows, lighting, etc.), using this as a workflow: viewtopic.php?f=3&t=1192.

Daz Studio is advertised as a 3D scene creation and rendering application, whereas Stable Diffusion can generate stunning image quality. Why not combine the two into a single workflow? At the end of the day, Stable Diffusion (and especially Flux) is far superior at coloring, texturing, lighting etc. than anything a 3D model can produce. That excludes post-production in an image software package like Photoshop. But if one is going to use Photoshop to finish an image, one may as well use ComfyUI instead. It's just a different software package. However, to get an AI model to understand what kind of a scene one wants, (ie. girl on the left, guy on the right, initial color scheme), an image is far superior to a text prompt. This is where 3D scene creation excels.

So far, ironically the hardest element in a whipping scene for me hasn't been the person receiving the whipping (download/file.php?id=74273 and download/file.php?id=77119 for some results) but the person dishing it out. The soldier swinging the whip from this image was the best result I could get from a model that recognizes "whipping, holding a whip" as input: download/file.php?id=76099&mode=view. Not terrible, but not exactly very convincing either. Whipping focused LoRA's exist, but so far a model's understanding of both the act of swinging a whip and the visual representation of the whip itself is surprisingly rudimentary.

[doe.1971]

I don't have any 3D skills, but if I did, I would use something like Daz 3D to create my scene, then pass the image to a Stable Diffusion model for the rest (texturing, shadows, lighting, etc.), using this as a workflow: viewtopic.php?f=3&t=1192.

I tried a few attempts of Poser (don't use DAZ) passed to SD as prompts and the results were ... lacking. Probably something I don't do right. Or the models that fit in 8GB are only limited to classic porn. Asking for poses more BDSM will invariably end up looking odd, despite what the graphic and verbal prompt asks.


But I agree that the future is very much like a crude drawing (as Jucundus did) or a rough 3D render used as a prompt for AI graphics. Probably by people much younger than me and maybe a decade in the future. For now, the VRAM is scarce and expensive, the models are sadly generic porn with little for pervs like us so yeah, it's a bit of a waiting game. But I can easily imagine a future where 128 GB VRAM is standard, there are specialized models for all the possible tastes and the interface will be mostly gesture based or vocally commanded.

But we are not there and gotta be realistic on what can be done.

[xs70]

I tried a few attempts of Poser (don't use DAZ) passed to SD as prompts and the results were ... lacking. Probably something I don't do right.

The trick is to change the parts that don't look right until everything does. Like how Mr. Hyde undresses these chicks on old pulp book covers.

[doe.1971]

The trick is to change the parts that don't look right until everything does. Like how Mr. Hyde undresses these chicks on old pulp book covers.

Still, it doesn't address the core issue, namely the lack of a large BDSM trained model (LORAs can help on particular cases but one needs a properly trained model to do something in this field. Mr Hyde use porn replacements, something that's abundantly available out there.

For example, try as you might, you'll never get a proper parrot perch position. Or a strappado. You can make a LORA but that's an add-on that's only good for that particular position. If I want a crucifixion next, I'm stuck again.

The technology is too new and it only caters to mainstream porn users so far. LORAs can help in punctual situations but they are not a replacement for a proper SM trained full model. As I said, in the future things will change, for sure.

[doe.1971]

From the days Yen spent being prisoner in Nilfgaard. Do comment if you like it.

[yyy02]

From the days Yen spent being prisoner in Nilfgaard. Do comment if you like it.


yen and voorhis v2.jpg

An unsolicited opinion from left field, but I love the subject, the position, and especially the way you did the lighting!

And the little glint of light from her toenails... ACE!

[doe.1971]

An unsolicited opinion from left field, but I love the subject, the position, and especially the way you did the lighting!

And the little glint of light from her toenails... ACE!

Merci, mon ami cubique!