Відео

Thank you for sharing your experience.

Because this workflow is simplified and does not use the diffuser node, ensure that the model file you are attempting to load is compatible with the version or UNET you are using.

@@goshniiAI can you help me using anydisck

It is nice to read from you, and you are very welcome.

I'm glad you found it easy to follow, you are most welcome.

Thank you so much! Cheers to you all the way in Estonia! 🌍

yeah there is a new one available

You're welcome, and that's a good observation. I'm using the older PulId model here because it’s highly compatible with FLUX. especially when dealing with complex inputs. However, the newer model offers exciting improvements!

You are right, you'd like to try it, and feel free to experiment as well

Thank you so much for your amazing comment! Tudo de bom para você também, meu amigo!

💚💚

Thank you for stopping by, and welcome to the family. I appreciate your encouraging feedback.

You are welcome. I'm glad it was helpful!

You are very welcome. thank you for stopping by.

I am using a 3060RTX with a 12GB VRAM.

It is encouraging to hear that. You are most welcome, and thank you.

uplifting. I appreciate you being here.

The workflow is available in this description. ua-cam.com/video/lbtQEtyvHps/v-deo.htmlsi=Y9suAe4030wuDdVO

-Sometimes ComfyUI struggles with incomplete filenames. Ensure the model file has a standard name and includes the correct extension (.ckpt or .safetensors). -If you’re using a FLUX model, double-check it matches the type required for the workflow. Mismatched models might cause this error. -then make sure the model file is saved in the correct folder. It should typically be in the directory of your ComfyUI installation. After checking the above, restart ComfyUI to allow it to rescan for available models.

Thank you lotss!

You are awesome for sharing your thoughts. you are most welcome

When everything stops after the AIO Aux Preprocessor group, it usually points to a missing connection or node dependency in your workflow. Open the ComfyUI console window and look for error messages. Sometimes a missing or mismatched model or node will throw an error there, giving you a clear idea of what’s wrong. If you’re using a custom node or model, ensure it’s compatible with FLUX and ControlNet. Sometimes older nodes can cause hiccups.

After installing custom nodes, not updating dependencies can sometimes cause compatibility issues, especially if the new node relies on updated libraries or scripts. so always click "Update All" in the ComfyUI manager. This ensures all dependencies for your nodes are correctly installed and compatible with your setup.

you’ve got some serious power under the hood, so the issue might not be related to the hardware. -Try lowering the resolution or batch size to see if it stabilizes. -Ensure all nodes in your workflow are compatible with FLUX and aren’t clashing. Sometimes reloading the model or rebuilding the workflow from scratch can help. -Double-check that there aren’t any corrupted or incomplete model files in your FLUX directory. -Make sure your GPU drivers and dependencies (like PyTorch, CUDA, etc.) are up-to-date and aligned with your system specs. You’ve got the hardware for greatness-I am hoping these will be helpful.

@@goshniiAI The problem is that I recently downloaded Comfy just to try it out, as I had previously used Automatic1111. That is, I had a fresh installation, following the instructions for Comfy. At first, I also thought that maybe I had chosen a too high resolution, so I gradually lowered it to 512×512. However, that didn't solve the issue. I have no idea what the problem could be, because it seems to stop working without any errors. It just stops after displaying the prompt, I have to start Comfy all over again..

thank you for taking the time. i appreciate your contribution.

Thank you for being here and joining the family. I appreciate your feedback and for bringing that to my awareness.

Thank you for providing feedback and suggestions. I have been working on including Pulid for quite some time now. I hope to overcome the challenges and share the process with everyone.

I have not been able to confirm whether A1111 is still operational.

Thank you for the feedback you got FLUX is running smoothly. Flux currently operates on a single GPU because ComfyUI does not support multi-GPU processing for workflows like this. However, some advanced setups, such as model splitting or custom modifications, may use multiple GPUs, but they require additional settings and may not be officially supported by Flux yet.

You should load the depth image into the load image node the same way but without using the preprocessor node. Hopefully, I will upload a video about that in the next post.

you could find the video here for ControlNet and redux helpful

Yes, that is still possible. sorry for not demonstrating it. By watching an image-to-image tutorial, you can connect your background to work with the workflow.

Thank you for the encouragement. We welcome you to the family.

@@goshniiAI 🙌

💚💚💜

Thank you, Lots! You are weolcome

Using the depth Lora, as I showed you, can help you avoid the large file size. The results remain the same.

Welcome to the team, and I am truly glad that my method is useful to you. It is encouraging to read that you value not only the what, but also the why of using these tools.

Yes, you can use the PULID Custum node FLUX for that. I have not been able to get PULID to function on my system with ComfoyyUI yet. you can view this reference to guide you: ua-cam.com/video/Uls_jXy9RuU/v-deo.htmlsi=qNoYR0xjk_A3COhB.

Rendering time can vary a lot depending on the complexity of the workflow and hardware capabilities. On a Mac with an M3 processor, ComfyUI might run slower since it’s designed more for NVIDIA GPUs. Flux models may also perform more slowly if you are just starting out. If you're considering switching to a Windows-based system. 1. An RTX 3070 or 4070 GPU with at least 16GB of RAM will give you a speed boost. 2. If budget isn’t an issue, go for an RTX 3080/4080 or 3090/4090 laptop with 24GB VRAM or higher. Also, pairing it with an SSD will improve overall responsiveness. After using flux for a while and setting, my generations took at least five minutes.

@@goshniiAI Thank you very much for your response!

Hello, here are a few suggestions that may be helpful. 1. After installing the node, restart ComfyUI completely to ensure it recognizes the new files 2. Double-check that the files are placed in the custom_nodes directory inside your ComfyUI installation. (IPADAPTER PLUS) 3. If you’ve already done that, the files might be outdated or incomplete. Run the update for the latest version inside the managerr.

You're not alone, once you get the hang of it, it’s incredibly powerful. I am glad the videos are helping to make it feel less scary. The warning in the "load checkpoint" section usually means that the checkpoint or model file is either missing or hasn’t been properly linked to ComfyUI. Try these few suggestions. 1.Download the Flux checkpoint model. The video here can guide you. ua-cam.com/video/TWSFej_S_bY/v-deo.htmlsi=oRN2j7ArRMTq9XRA 2. Make sure the model file you're trying to load is in the correct directory: the /models/checkpoints folder. 3. Double-check that the model is in a compatible format, like .safetensors 4. After placing the model file in the right folder, restart ComfyUI to let it detect the new files.

@@goshniiAI you are the king thank you so much

@@metinozkantar8789You are most welcome.

@@goshniiAI Will there be a version with detailed face details? Such an educational video would be useful to people as we can see their facial expressions from a few poses.

You're welcome! Glad it was helpful.

LOVE!

Thank you for being here and taking the time to share your thoughts! We welcome you to the family.

You are most welcome.