Update: Hopefully 1.2 will work out better with more aesthetic and accurate gens counting the previous problems are mostly gone, still not the ideal version, it was a hard choice between the candidates so I had to balance out between unstable and featureless. There might be a problem where the first RNG roll after you change the prompt will be inaccurate. Everything else should be good to go and I'll keep the 3D styles coming with it, thank you for all the downloads and likes!
This model still works pretty similar to base version except the 2D content is culled by default and some baked characters will behave better. It's not 100% round but it's already quite the leap, still need some tuning that'll be made on version 1.2 which is are the problems you might find now: inpaint artifacts, poor upscaling with some samplers and low style flexibility at score_9.
Recommended sampler/scheduler: DPM, Euler A /SGM Uniform, Karras
Recommended CFG: 5
Recommended upscalers: Latent /Tiled Upscalers
Should be more stable, more compatible with LoRAs while maintaing style and keeping all the capabilities of the base version/PONY V6.
描述:
训练词语:
名称: dogepoint3DPDXL_dogepony3DV10.safetensors
大小 (KB): 6775430
类型: Model
Pickle 扫描结果: Success
Pickle 扫描信息: No Pickle imports
病毒扫描结果: Success
