Update 9-28-25: I added a (much smaller) V3. This one uses a dim/alpha of 16/8 (so 109 MB) this time, I changed the tagging, and removed a handful of images from the dataset. It probably won't look as good as V2, but it'll be a bit more flexible and be 1/4th the size.
Update 6-27-25: I added a (much improved) Illustrious version 2. I made it a THICC 435 MB LoRA (i.e., Dim 64, Alpha 32) to get it as close as possible to the movie.
A commission! I've never made a 3D model before, but since Tifa/Final Fantasy VII is an honorary anime anyway...yeah. Overall, I feel Stable Diffusion is better for 2D characters over 3D (3D CGI often can look uncanny).
Anyway, I freaking love Tifa! So it's great that I was given an excuse to create a model of her! As the title suggests, this is based on her Advent Children design. I did both a Pony version and an Illustrious version. Naturally, the Illustrious version will produce more illustration-style images; however, since it has a CG-flair to it, I'm labeling it as 2.5D.
Obligatory self-promotion: consider supporting or commissioning me at https://www.patreon.com/MetalChromeX so I can continue making cool crap for you all to enjoy for free!
Description
FAQ
Comments (8)
5090 are a scam, dont fall for it, taking 40-30 gen is more than enought, corpo greed are crazy and your doing well without a 50. + I dont think they resolve the problem of the burning pin cuz they overheat.
Oh, I actually just bought a 5080 lol (got it just a bit over MSRP, so not too bad). That was an old carryover description from the V1 model; I'll update that in a sec. I don't think the 5090 would have been a good value for the cost, especially since they're like $1000 over MSRP most of the time. Going from a 2070 Super to a 5080 has been worth the updgrade/investment, in my opinion.
@metalchromex 1000 bucks for GPU? Im in EU, i upgrade from 1070 ti for 2080ti 11 Gg for the equivalent of 360 Euro/$. The price are beyond bs for GPU these days, its a good investment of course if you go full throttle with your GPU but damn, the price hurt
@metalchromex Congrats in achieving your goals, Chrome!
Since a long time that you were trying to get this new GPU upgrade. And you definitely derserve it!
Have a great day!
@metalchromex Don't listen to these cretons. 5090 is a great card. Its not the ad promises and marketing that makes it great, its the memory and memory spe*ed. 32 GB and 70% faster memory bandwith than 4090, clearly makes it the top racing horse for AI applications. For Gaming it amy be a bit over the top at current prices, but for AI there are not many options except for the Workstation GPUs starting at 6000 EUR
You should ask around with your other fellow LoRA creators how to make good LoRA at acceptable size. 460 MB is just way overkill. Some of the good LoRA creators have SDXL/PDXL/ILXL LoRas at 26-52 MB, 1/8 of that.
This is a major issue for people who have a lot of LoRAs
Funny you mention this; I was literally just about to test a smaller sized version. I'd just heard that 3D-based models work better with a 64/32 dimm/alpha, but yeah it doesn't appears to make that much of a difference from what I've seen. I'll plan to re-do a few of the older 3D-based LoRA that were probably a bit too large in size.
@metalchromex Much apperciated. I got 1.5 TB of LoRAs, 200-400 MB LoRA really hurt my drive much haha (I know, I'm a hoarder...)
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.