I post models on multiple sites, including exclusive models. For new model releases, bounties, and more, follow me on: <a target="_blank" rel="ugc" href="https://x.com/DarkInfinityArt">X (Twitter)</a> or <a target="_blank" rel="ugc" href="https://x.com/DarkInfinityArt">Instagram</a>. Also, check out <a target="_blank" rel="ugc" href="https://darkinfinityart.blogspot.com/">LoRA Lore</a> for information on how to train your own models. More info &amp; links: <a target="_blank" rel="ugc" href="http://www.darkinfinity.art">www.darkinfinity.art</a>This model card is for pre-release versions of new LoRAs that I plan to post soon. I will post bounties for the best AI artists to help generate showcase images that show off the new model's full potential.If you are participating in a bounty, please do not post your bounty images to this model card or leave reviews for the model. This is not an attempt to solicit reviews or encourage image posts on this model. To ensure the Civitai TOS is followed (and to avoid duplicate models), the version of the LoRA posted here will be deleted once the bounty is over. The LoRA will be reuploaded on its own model card with the bounty-winning images as its showcase.Any current bounties will be listed below. Follow the link to the Civitai bounty page to learn more.<h3 id="scrollbound-lore-bounty">Scrollbound Lore Bounty</h3>Scrollbound Lore complements <a target="_blank" rel="ugc" href="https://civarchive.com/models/1456794/ink-and-lore">Forgotten Tales in the Ink &amp; Lore LoRA series</a>. If you run low on buzz,<a target="_blank" rel="ugc" href="https://www.mage.space/play/7aabe29175404744bde5a31dc033185c"> you can run the LoRA for free on Mage</a> until the bounty ends.The <a rel="ugc" href="https://civarchive.com/bounties/9830">bounty </a>has ended, but the new model is available on the <a target="_blank" rel="ugc" href="https://civarchive.com/models/1456794/ink-and-lore">Ink &amp; Lore LoRA series model card. </a>This LoRA generates illustrations in a storybook-meets-woodblock style, strongly leaning toward Asian myths and legends. It is trained on a mix of traditional asian art styles from Japan, China, and Vietnam, and more modern depictions of various mythological and legendary characters, spirits, and creatures. Subjects are depicted with clean outlines, flat expressive shading, and balanced, muted color palettes reminiscent of the traditional Japanese and East Asian woodblock prints, and backgrounds often feature natural or symbolic elements — moons, mountains, temples, water, or spirits — creating the feel of mythic scenes frozen in time.Use the trigger word, “distrnmyth” along with descriptions of animals, spirits, yokai, and other elements of magic and folklore. The LoRA was trained on Flux.1 Krea, but works on Flux.1 Dev as well.<h3 id="aqua-velvet-krea-lora-beta">Aqua Velvet Krea LoRA BETA</h3>A hand-drawn manga/anime glamour look that favors soft bodies with flattering proportions. Faces are soft with large expressive eyes; lines are clean and confident, and details are minimal, but expressive. Poses lean close and a bit foreshortened, with subtle highlights and shading to suggest soft skin and curve without over-detailing.This LoRA is trained to be used with the Flux.1 Krea model, but it should work equally well with the Flux.1 Dev model. The differences in the base model will result in slightly different colors and a good deal of difference in composition and prompt understanding. Using either base model, trigger with "diqvlvt" and ask for anime or manga art. Here’s a sample prompt template:<ul><li>diqvlvt style manga art, {prompt}</li></ul>Special thanks to reddit user insanebuddy09 who provided much of the dataset for this model.<h3 id="dungeons-and-dreamscapes-bounty">Dungeons and Dreamscapes Bounty</h3><ul><li>The <a target="_blank" rel="ugc" href="https://civarchive.com/bounties/9208">#bounty# (https://civarchive.com/bounties/9208)</a> has ended, and this model is now published as <a target="_blank" rel="ugc" href="https://civarchive.com/models/1794277?modelVersionId=2096879">Dungeons &amp; Dreamscapes v2.1 for Flux. </a>You can also run it on Mage.Space as the <a target="_blank" rel="ugc" href="https://www.mage.space/play/d2679449feb143f79458f641903bfee8">Dungeons and Dreamscapes Style concept</a>.</li></ul>Dungeons and Dreamscapes specializes in generating atmospheric scenes within the fantasy genre. Trained on a diverse set of highly detailed, narrative-driven captions, it excels at producing moody medieval landscapes, mysterious ruins, ancient forests, and evocative character portraits.The model’s outputs are distinguished by their cinematic lighting, muted earth tones, and richly textured digital painting style.Use the ridiculously long trigger word, “didngnsnddrmscps” along with tokens like digital fantasy art, painterly painting style, moody atmosphere, chiaroscuro lighting, muted earth tones, highly detailed, atmospheric. Here’s a sample prompt template:<ul><li>didngnsnddrmscps, digital fantasy art, {prompt}, painterly painting style, moody atmosphere, chiaroscuro lighting, muted earth tones.</li></ul>Special thanks to reddit user XoxoH123 who provide much of the dataset for this model.<h3 id="silk-and-shade-lora-beta">Silk &amp; Shade LoRA BETA</h3>Note: The new version of the new models output after it was fully developed and is available on<a target="_blank" rel="ugc" href="https://www.mage.space/play/85ae9aada8d4410e996966cade7fc125"> Mage.Space as the concept Velvet &amp; Shadow concept.</a> Some of its output:<img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/5a0dd4ce-437a-49a9-be5a-d72a1b40e2ad/width=525/5a0dd4ce-437a-49a9-be5a-d72a1b40e2ad.jpeg" />New model I've been working on but not quite happy with yet. Feedback would be appreciated and help me complete it. Though it's based on dark manga-style artwork, it has some other imagery mixed in. Images are supposed to be dark and shadowy, but you should also be able to prompt for interesting lighting, high contrast, and a few other artistic effects.<ul><li>Use LoRA trigger phrase "a dinctrlns style photograph" or "a dinctrlns style drawing" -- it should work for other mediums as well.</li><li>Lots of words related to darkness and shadows should enhance the LoRAs effect.</li><li>Standard Flux settings should work -- Euler Simple, 20 steps, Flux.1 Dev base model.</li></ul><h3 id="baroque-macabre-style-lora-bounty">Baroque Macabre Style LoRA BOUNTY</h3><ul><li>The <a target="_blank" rel="ugc" href="https://civarchive.com/bounties/8303">bounty </a>has ended, and this model is now published as <a target="_blank" rel="ugc" href="https://civarchive.com/models/1647182">Divine Decay: Macabre Baroque style LoRA</a>.</li></ul>Baroque Macabre Style is a high-contrast, engraving LoRA that emulates intricate linework and dense sketch textures. Common motifs include twisted trees, animal skulls, gothic architecture, and starry night skies. Ideal for generating monochrome illustrations with dark fantasy or occult themes, this LoRA responds well to tokens related to bones, skulls, demons, monsters, and other things that go 'bump' in the night. The model is also a big fan of fancy scrollwork, twisting patterns, and decorative knotwork.<ul><li>Use the trigger word "dipnmbrx" along with a style, as in "a dipnmbrx style illustration"</li><li>Many styles work, and the trigger and LoRA together make everything from baroque-style engravings to dark fantasy photorealistic. It has been tested with sketches, illustrations, paintings, drawings, engravings, etchings, photographs, and line art.</li></ul><h3 id="classic-oil-painting-lora-bounty">Classic Oil Painting LoRA BOUNTY</h3><ul><li>This is a model I made for personal use, but it turned out well, so I'm going to try to post it. It has an Easter egg you may find while generating, so have fun with this one.</li><li>Use LoRA trigger phrase "a dilpntng style oil painting"</li><li>Standard Flux settings should work -- Euler Simple, 20 steps, Flux.1 Dev base model.</li><li>The <a target="_blank" rel="ugc" href="https://civarchive.com/bounties/7790">bounty </a>has ended, and this model is now published as <a target="_blank" rel="ugc" href="https://civarchive.com/models/1482404">Academic Old‑Master Oils</a>.</li></ul><h3 id="folk-art-style-lora-bounty">Folk Art Style LoRA BOUNTY</h3><ul><li>This style is part of a new series I'm creating that produces art reminiscent of folk art or fairytale illustrations. This one reminds me of early Nordic artwork, but that may just be me.</li><li>Use LoRA trigger word "dijldflklr"</li><li>Standard Flux settings should work -- Euler Simple, 20 steps, Flux.1 Dev base model.</li><li>The <a target="_blank" rel="ugc" href="https://civarchive.com/bounties/7703">bounty</a> has ended, and this model is now published as <a target="_blank" rel="ugc" href="https://civarchive.com/models/1456794">Ink &amp; Lore: Forgotten Tales</a> (or will be soon).</li></ul><h3 id="concept-art-sketch-style-lora-bounty">Concept Art Sketch Style LoRA BOUNTY</h3><ul><li>This style was trained on concept art from video games and with the trigger word “digrngbrsh” but like most Flux LoRAs, the trigger word may not be necessary as long as you include associated style tokens like “a sketch of” when you use the LoRA.</li><li>Images should have a rough sketch appearance, similar to the examples on the model card.</li><li>Standard Flux settings should work -- Euler Simple, 20 steps, Flux.1 Dev base model.</li><li>The <a target="_blank" rel="ugc" href="https://civarchive.com/bounties/7579">bounty</a> has ended, and this model is now published as <a target="_blank" rel="ugc" href="https://civarchive.com/models/1433827">Sketch Pad: a Concept Art LoRA</a>.</li></ul>

Dark Infinity's Pre-release Models

A hand-drawn manga/anime glamour look that favors soft bodies with flattering proportions. Faces are soft with large expressive eyes; lines are clean and confident, and details are minimal, but expressive. Poses lean close and a bit foreshortened, with subtle highlights and shading to suggest soft skin and curve without over-detailing.

aqua-velvet-krea_v10-000013.safetensors

Aqua Velvet Beta v1.0

Files with SHA256: 0f3dc54c850b5ebf71a3ee1cd9a536b3e3bc72021b8f0d688fe898040c27abf2