6G VRAM, recommended to use FP8 model
After my testing, local 6G VRAM can directly generate 2048*2048 resolution images using an fp8 model (note to update to the latest version of comfyui)
（The model download links are all added in my workflow.）
Click to try it out directly:https://www.runninghub.ai/ai-detail/1993973002423435266/?inviteCode=rh-v1315
Video introduction:
More workflow parameter settings, please refer to my settings on rh:https://www.runninghub.ai/post/1993890080389611521/?inviteCode=rh-v1315
You can directly click the link above to try it, no need to download the model, click to run the workflow online. If the results are good, you can deploy it to your local environment.
Sign up via this link claim 1000 RH Coins and generate tons of images/videos for FREE!
My AI models/apps are on RH now! Try them out & support my work — every click counts. Thank you!
Thank you all for your support. Your support is the motivation for me to update more interesting and useful workflows. Thank you so much!!!
Description
FAQ
Comments (13)
"I have a question. My 12GB GPU seems unable to run qwen3-4b; I keep getting an 'OutOfMemoryError: Allocation on device' every time."
My 4060 ti laptop can afford it.(just 8GB)
Maybe you should update comfyui. I also encountered this issue at first, but after updating to the latest version of comfyui, it ran normally
@Zovetry Thanks
It worked! I can use it now. Thanks! Who knew it just needed an update?
why my rtx2060 make 1 image(1024*1024) need 200+ s? fp8/bf16
It might be that 20 series graphics cards don't support certain functions. On my local machine, using a 3060 6G to generate 1920*1024 takes only about 100 seconds
float节点无法安装，comfyui更新到最新也无法安装该节点
In fact, this node is not necessary; you can delete this node. Then you can directly change the noise reduction value in the sampler.
请问这是什么节点呢FlowMatchEulerDiscreteScheduler (Custom)
这是zimage官方使用的一个调度器
FlowMatch Euler Discrete Scheduler (Custom)这个节点报错
You just need to delete the node and reload it, the error is due to an interface issue caused by the node update