Tag: AI
-
Stable Diffusion [End May Miscellany]
May was a quiet month for the blog as I concentrated on the (real world) setup for WittleWaifus and getting my graphic design processes down pat to streamline sticker production. I did manage to get in a bit of Stable’ing here and there, and so I present the second entry in my series of miscellaneous…
-
Stable Diffusion [VHS Style]
In my directionless prompting, I stumbled across a combination of terms whose output is that of girls posing for intimate photographs in locales packed with retro-tech, in a style that wildly varies between “sort of manga panel” and “sort of animation cell.”I’ll dub this “VHS Style” for now and fully intend to explore the concept…
-
Stable Diffusion [Spring Showers]
I combined my shitmix anime model with animelike (70:30 weight) and- since I just watched Weathering With You– ran a few pictures of girls in the rain as a test: They pictures came out… Fine. Animelike’s faces are a tad too expressionless for me, so I might need to remix things again with a heavier…
-
Stable Diffusion [Going Back]
Whoa! I hadn’t realized that it’d been so long since I last posted. Truth be told, I’ve been busy with the work of getting my sticker shop ready for launch and haven’t been doing much Stable’ing for fun. To get back into it, I loaded up the prompt I used to generate some of the…
-
Stable Diffusion [Animelike Anime]
Although gone from Civit.ai, the “animelike” models (2D and 2.5D) are still out there. While they have the benefit of producing pictures that easily be stills from any number of modern shows, I’ve always felt that the output is just a little too clinical. However, that’s probably in keeping with modern anime anyway, what with…
-
Stable Diffusion [Asuka’d]
I started up the Stable webui only to realize I had no direction or real idea as to what I wanted to generate. I jumped from model to model, adding and removing LoRAs and generally testing things out in an extremely unscientific way. In the end, I ended up on a prompt for best girl…
-
Stable Diffusion [Ghibli Beach Day]
Some results from a Studio Ghibli-style LoRA, albeit with the weight turned down considerably (~0.4) and combined with the lineart and watercolour LoRAs I was using before.
-
Stable Diffusion [Limited Palette Asuka]
Yeah, yeah. I know 22% of this gallery is full colour. Whatever, it’s my site.
-
Stable Diffusion [EVA Screencap Edition]
A few results generated whilst trying out the “Evangelion (1995) style” LoRA. I’m definitely going to be using this one a bunch in the near future.
-
Stable Diffusion 22 Apr 23
Another batch of sketchy pics. These are completely untouched (no inpainting, etc).
-
Stable Diffusion 19 Apr 23
A sort-of-sketchy, sort-of-painterly update today, with pictures generated by AnythingV4 in combination with a pair of Loras to handle the processing for the lineart and watercolour styles. Only the most egregious anatomical errors were sent to inpainting for correction, in keeping with the sketchy nature of the images.It’s interesting how- when the bot chose to…
-
Stable Diffusion 17 Apr 23
Continuing on with my wildcard experiment, I tweaked the clothing related cards to all-dress-specific items and added an emote/expression variable. In addition, I inadvertently found a pair of Loras that (with proper weights), give a sort of ink + watercolour result that I find rather pleasing.