when i use ltx-2.3-22b-distilled-fp8.safetensors ，i get very bad results；
python -m ltx_pipelines.distilled --quantization fp8-cast --distilled-checkpoint-path "E:/AI_model/LTX-2.3-fp8/ltx-2.3-22b-distilled-fp8.safetensors" --gemma-root "E:/AI_model/gemma-3-12b-it-qat-q4_0-unquantized" --prompt "A cinematic shot of a golden retriever playing in a field of flowers, golden hour lighting, 4k, highly detailed." --output-path "./outputs/my_video.mp4" --height 512 --width 768 --num-frames 121 --frame-rate 24 --spatial-upsampler-path "E:/AI_model/LTX-2.3/ltx-2.3-spatial-upscaler-x2-1.0.safetensors"
when i use ltx-2.3-22b-distilled-fp8.safetensors ，i get very bad results；
python -m ltx_pipelines.distilled --quantization fp8-cast --distilled-checkpoint-path "E:/AI_model/LTX-2.3-fp8/ltx-2.3-22b-distilled-fp8.safetensors" --gemma-root "E:/AI_model/gemma-3-12b-it-qat-q4_0-unquantized" --prompt "A cinematic shot of a golden retriever playing in a field of flowers, golden hour lighting, 4k, highly detailed." --output-path "./outputs/my_video.mp4" --height 512 --width 768 --num-frames 121 --frame-rate 24 --spatial-upsampler-path "E:/AI_model/LTX-2.3/ltx-2.3-spatial-upscaler-x2-1.0.safetensors"