Model is trained on 44 hand picked and hand tagged images, mostly illustrated works, others being 3d artwork. CivitAI only partially tagged the images, I added detailed tags for better prompting. This is a V1 model, meaning output will be accurate to what modern interpretations of what an allosaurus looks like, but it might be lacking in some areas. For Pony, I cannot attest to the quality, but the epoch versions in training came out much more coherent. <a target="_blank" rel="ugc" href="https://civarchive.com/models/503815?modelVersionId=1032006">Nova Furry XL V6.0 </a>may work better, but is untested by me. <a target="_blank" rel="ugc" href="https://civarchive.com/models/784543?modelVersionId=1028888">Nova Animal XL V4.0</a> is what was used for the sample pictures for the Pony LoRA. This LoRA is capable of generating NSFW content, tag negative prompts accordingly to avoid potential NSFW content. Tagging: Tagging works best with the following tags to accurately prompt an allosaurus, the tag "Allosaurus" is enough to trigger the LoRA. eye crest, (color) eye crest, (color) scales, (color) highlight, (color) underbelly, optional tags are: furry body, (color) fur, illustriousV2: V2 adds a bigger dataset, from 44 to 64. steps have been increased from 563 to 1560. added tag: realistic feral, (color) eyelids, PonyV2: V2 adds a bigger dataset, from 44 to 64. steps have been increased from 450 to 1248. added tag: realistic feral, (color) eyelids,

Allosaurus LoRA

Files with SHA256: a0212ed3beb93cc27ed9ba18514124e3a3047237f0aec8300fba9cf6f6f772ff