More new(ish) LORAs which may appeal to Tentaclii readers. LORAs being free plugins that ‘guide’ the AI images made with your local install of Stable Diffusion 1.5.

* New this week is the somewhat Solomon Kane-ish Witch Hunters, which by the look of it was also trained with some over-the-top corporate card-art. Still, you can stack LORAs, thus you could blend it with another such as James Purefoy (Solomon Kane) which was trained on movie stills.

* One I missed, from a year ago, CthulhuTECH. Every image, no matter what the subject, becomes somewhat Cthuloid. This one is a LYCORIS, but is used in the same way as a LORA. CthulhuTECH is very well documented, so make sure to click “Show more” and then save a .MHTML page alongside your download.

Also has a cthulhuTECH books add-on as a LORA.

* Looking at the CthulhuTECH maker’s other stuff, I see I also missed his Pulp Ladies – In Spaaaaace. Again, it’s a LYCORIS but could still be combined with a LORA to give the image output a more ‘aged pulp cover’ feel.

New on YouTube, “H.P. Lovecraft Poetry, Converted into Goth Alternative Rock by AI”.

Does Stable Diffusion 2.1 768 know what Lovecraft looks like, with a simple name prompt of ((H.P. Lovecraft)) ? After all the hoo-ha about celebrity pictures and name-tagging being removed from the training of SD 2.x? Well… yes, sort of. The results are not bad, considering how few good pictures there are of him. But not great.

Lovecraft’s long and earliest ‘cosmic’ poem “The Poe-et’s Nightmare” (1916), as an AI generated video adaptation. The project…

aims to decode the poem’s innate symbolic space and to project it into a coherent visual realm with a surreal aesthetics of a graphic antique manuscript, abstracting away from the described events. Eventually, what we observe is a dialogue between the written unseen and the unspeakable evident. The collection consists of 304 unique pieces, one per verse line. Each piece is a 23-second video loop with sound.