Search Results
7/22/2025, 11:39:21 PM
>>105992335
thanks for reposting.
more neta lumina testing... it struggles with a simple prompt that noob does a great job at. even after some prompt re-engineering. bad hands and anatomy.
https://files.catbox.moe/ihmwnt.png
Anyone want to criticize this prompt? Clearly, looking at the samples here https://nieta-art.feishu.cn/wiki/RY3GwpT59icIQlkWXEfcCqIMnQd, the model is capable of better. but it seems like you have to prompt it a special autistic way to get it to work, and I'm failing.
>>105991800
this benchmark is outdated and people should stop posting it. the 7900 XTX is around a 3090 in perf, not a 4060 Ti lmao. I'm getting 1it/s at standard resolutions in Lumina rn, with no optimizations (eg FlashAttention) enabled. this pic took 30 seconds for 30 steps.
thanks for reposting.
more neta lumina testing... it struggles with a simple prompt that noob does a great job at. even after some prompt re-engineering. bad hands and anatomy.
https://files.catbox.moe/ihmwnt.png
Anyone want to criticize this prompt? Clearly, looking at the samples here https://nieta-art.feishu.cn/wiki/RY3GwpT59icIQlkWXEfcCqIMnQd, the model is capable of better. but it seems like you have to prompt it a special autistic way to get it to work, and I'm failing.
>>105991800
this benchmark is outdated and people should stop posting it. the 7900 XTX is around a 3090 in perf, not a 4060 Ti lmao. I'm getting 1it/s at standard resolutions in Lumina rn, with no optimizations (eg FlashAttention) enabled. this pic took 30 seconds for 30 steps.
Page 1