Hey everyone, since so many of you asked I added the link of the workflow in the description. With that being said, the purpose of this step by step guide was to show you how to create it from scratch as this video is part of the Comfyui 101 series of videos. Do yourself a favor and make the effort to build it yourself as I put a lot of time into this video. So go ahead and download the workflow but use it to "LEARN" how to create it. Don't just take the workflow, use it and put it aside, you'll learn nothing, use it as another resource to understand Comfyui. Those of you who already follow me know that I give my workflows out freely, so it's not about that. I'm on a mission to make Comfyui more understandable for the average person but if you don't put in the time to learn it, then there is nothing I can do at that point.
I know this one is a long one but there was a lot to go over. I also wanted to share some of these tips and tricks. There is a part 2 coming soon, otherwise this video would've been wwwaaayyyy too long! As always, you can support my channel by liking, commenting and watching the video through! Much love M&M fam!
I appreciate the positive vibes my man! I had to cut out a lot of content as the video was already longer than I would've liked but even more reason for a part 2 for upscaling and we can dive in deeper. BTW off topic....do you really play bass? 😁
Thank you. But I just showed how to build it from scratch! Best way to learn is to build it yourself. I added a link in the description to the workflow but make it a point to build it yourself otherwise you will never learn.
@@SandhyaMeenakshi-t3h you can install the node "ComfyUI SaveAS" and connect it as you normally would at the end of the workflow. This node allows you to save your image in the most common image extensions. Just keep in mind jpg is more compressed so there is a loss in quality. png is much better.
Thank you so much for this series sir. One question, in your personal workspace that you shown at the end, where are all the pipelines? Are they hidden? How do we do it ?
You're welcome! Glad you are enjoying the series. Yes they are hidden as shown here 25:23 You just have to toggle visibility off. Bottom right corner. You can also do it in settings>lite graph>graph and also change the look of the connections to spline, straight, linear or hide.
@@MonzonMedia Thank you for the quick response. I missed that part while experimenting with my own. Request you to add a workflow for the Flux gguf q8 if possible.
My friend if you've been following this series you would already know how to do that on your own. That's why I only mentioned it and didn't show how to do it. You just have to replace the load checkpoint node with the GGUF loader, add GGUF dual clip loader, VAE loader and connect the nodes accordingly. I go over it in this video here. ruclips.net/video/DBoms31-dVg/видео.htmlsi=cWPEnXN8PciG3I73 Same concept applies. And if that video is still confusing than I suggest you start from the beginning with this video. ruclips.net/video/Yk8aS233HP0/видео.htmlsi=a4a7SiNRx67leXLm
@@MonzonMedia alright I'll go through it again. I have been following the series closely, while using gguf I was having some issues that were confusing me, hence I thought I'll ask you. Thank you for helping out, apologies for bothering you :(
@@rajatshrivastav450 Not a bother at all. I'm glad you ask so that's why I directed you to that video. I wasn't sure if you were following the other videos but I did cover using GGUF models in that video I linked. It's towards the end if you don't need to know the other stuff. As I said, the same concept applies and you just have to do the same thing on this workflow. Very simple. If you are still having issues let me know. I'm always happy to help! 👍
Yes you can watch it here ruclips.net/video/SXt9dbtHy6U/видео.htmlsi=FtK2gSpynKH_NMFl make sure to watch the other videos otherwise you won’t understand the rest.
Hahahaha! I can't get away from upscaling videos this morning. I've only been awake 2 hours and this is my third with a focus on it already. It's always a time vs quality trade-off with upscaling. Do you do it with the initial generation with the added time cost to every generation, or after where you can pick which ones to upscale. I prefer the latter so this will be very useful. One question, can this be adapted to do batch operations, for example all files in a folder? If it can, that may be something for a follow on video. 🙂
Hey bud always a pleasure to hear from you. I usually only upscale as a separate process and as mentioned it's the last thing I do. I have created workflows that have a pipeline that ends with an automated upscale and you can just bypass it when not used. However because of my system limitations I choose not to do it that way. As for the batch method, I believe so. I recall seeing another video on upscaling awhile ago (it was for SDXL) where the user was using a particular node where you can import a folder of images.....I don't remember off the top of my head but I'll make a note to check it out.
@@MonzonMedia I figured Comfy would be able to do batches but I'm not familiar enough with it to know and Forge has it built in so I use that. I tend to save mine up, then when I have a few, do them all in one go over lunch or something. That was mainly with SD1.5 as I wanted higher resolutions but with SDXL and Flux I can get there straight away.
@@Elwaves2925 Totally hear ya. If I were to do batch runs then yes I'd do it on Forge since it's so much easier. I'm staying hopeful that the devs will come back to it soon. I really do like using Forge. Since it's a clone of A1111 it just feels natural to use since A1111 was the first local platform I learned.
But I just showed how to build it from scratch! Best way to learn is to build it yourself. I added a link in the description to the workflow but make it a point to build it yourself otherwise you will never learn.
For images invokeai hands down, it's just easier and fits my workflow. Comfyui is great for all purpose things such as video and of course if I want a custom workflow I can create it in comyfui. I would always recommend invokeai to anyone that wants a good stable platform.
This simple workflow is designed specifically for AI-generated images and is not suitable for real photographs, especially older ones. For real-life photos, additional steps are required to reduce compression artifacts and denoise the image. You also need to incorporate extra stages for ControlNet; otherwise, the person in the photo may not retain their original appearance after upscaling. Additionally, if the source photo is too large, processing it as a single piece can lead to out-of-memory (OOM) errors. If you use tiling to avoid this, it may introduce seam artifacts across the image.
True but how can one learn how to do these things without the knowing and understanding the basics first? Which is the purpose of this video and part of the Comfyui series on this channel.
Hey everyone, since so many of you asked I added the link of the workflow in the description. With that being said, the purpose of this step by step guide was to show you how to create it from scratch as this video is part of the Comfyui 101 series of videos. Do yourself a favor and make the effort to build it yourself as I put a lot of time into this video. So go ahead and download the workflow but use it to "LEARN" how to create it. Don't just take the workflow, use it and put it aside, you'll learn nothing, use it as another resource to understand Comfyui.
Those of you who already follow me know that I give my workflows out freely, so it's not about that. I'm on a mission to make Comfyui more understandable for the average person but if you don't put in the time to learn it, then there is nothing I can do at that point.
I know this one is a long one but there was a lot to go over. I also wanted to share some of these tips and tricks. There is a part 2 coming soon, otherwise this video would've been wwwaaayyyy too long! As always, you can support my channel by liking, commenting and watching the video through! Much love M&M fam!
👋
@ 👊🏼👋🏽
Good stuff. Your OCD desire for neatness in the workflow speaks to me 😂
@@zipzapson8602 LMAO! Someone else get's it!!!! hahaha! 🤣
You're the best man!!! The way you explain things is unmatched. Keep making more videos like this. ❤❤❤
Thanks man! I really appreciate it. I have fun playing around with comfyui and I'll be sure to keep making more!
You never stop learning. Ty
This is true! I'm addicted to learning! 🙌
Another excellent video that continues your pattern of providing an optimal mix of the why with the how-to and the practical.
I appreciate the positive vibes my man! I had to cut out a lot of content as the video was already longer than I would've liked but even more reason for a part 2 for upscaling and we can dive in deeper. BTW off topic....do you really play bass? 😁
Excellent tutorial. Can you please provide the link to workflow?
Thank you. But I just showed how to build it from scratch! Best way to learn is to build it yourself. I added a link in the description to the workflow but make it a point to build it yourself otherwise you will never learn.
Love your videos, very helpful. Awesome tip about creating group nodes.
Very handy little trick that makes things so much cleaner. I appreciate you!
I was waiting for this from you ❤️❤️❤️🙏🙏🙏
Hope you enjoy it! 🙌🏼😊
@ it was great as usual. Just one thing - how do u save the image generated as jpeg instead of png by default? 🙏🙏🙏
@@SandhyaMeenakshi-t3h you can install the node "ComfyUI SaveAS" and connect it as you normally would at the end of the workflow. This node allows you to save your image in the most common image extensions. Just keep in mind jpg is more compressed so there is a loss in quality. png is much better.
Thank you so much for this tut!
No problem, you are very welcome! 😄
Thank You 🤗❤
You’re welcome 😊
Great one! Thanks!
You bet! Appreciate the support.
Yes, please more in-painting, and about upscaling in this video, does it upscale the image for examle X2, or it only enhances the details
Yes it does both upscale and enhance details as shown in the video. Inpainting to come.
Thank you so much for this series sir. One question, in your personal workspace that you shown at the end, where are all the pipelines? Are they hidden? How do we do it ?
You're welcome! Glad you are enjoying the series. Yes they are hidden as shown here 25:23 You just have to toggle visibility off. Bottom right corner. You can also do it in settings>lite graph>graph and also change the look of the connections to spline, straight, linear or hide.
@@MonzonMedia Thank you for the quick response. I missed that part while experimenting with my own. Request you to add a workflow for the Flux gguf q8 if possible.
My friend if you've been following this series you would already know how to do that on your own. That's why I only mentioned it and didn't show how to do it. You just have to replace the load checkpoint node with the GGUF loader, add GGUF dual clip loader, VAE loader and connect the nodes accordingly. I go over it in this video here. ruclips.net/video/DBoms31-dVg/видео.htmlsi=cWPEnXN8PciG3I73 Same concept applies. And if that video is still confusing than I suggest you start from the beginning with this video. ruclips.net/video/Yk8aS233HP0/видео.htmlsi=a4a7SiNRx67leXLm
@@MonzonMedia alright I'll go through it again. I have been following the series closely, while using gguf I was having some issues that were confusing me, hence I thought I'll ask you. Thank you for helping out, apologies for bothering you :(
@@rajatshrivastav450 Not a bother at all. I'm glad you ask so that's why I directed you to that video. I wasn't sure if you were following the other videos but I did cover using GGUF models in that video I linked. It's towards the end if you don't need to know the other stuff. As I said, the same concept applies and you just have to do the same thing on this workflow. Very simple. If you are still having issues let me know. I'm always happy to help! 👍
can you do a video about ace++
I actually recorded something the day after it came out but I didn't like what I did so I decided to re-record it. Hopefully later this week.
Very cool video! Have you already shown it in ComfyUI IMG2IMG, or is one in the making?
Share the workflow please!
Yes you can watch it here ruclips.net/video/SXt9dbtHy6U/видео.htmlsi=FtK2gSpynKH_NMFl make sure to watch the other videos otherwise you won’t understand the rest.
I added a link in the description to the workflow but make it a point to build it yourself otherwise you will never learn. 👍
Hahahaha! I can't get away from upscaling videos this morning. I've only been awake 2 hours and this is my third with a focus on it already. It's always a time vs quality trade-off with upscaling. Do you do it with the initial generation with the added time cost to every generation, or after where you can pick which ones to upscale. I prefer the latter so this will be very useful.
One question, can this be adapted to do batch operations, for example all files in a folder? If it can, that may be something for a follow on video. 🙂
Hey bud always a pleasure to hear from you. I usually only upscale as a separate process and as mentioned it's the last thing I do. I have created workflows that have a pipeline that ends with an automated upscale and you can just bypass it when not used. However because of my system limitations I choose not to do it that way. As for the batch method, I believe so. I recall seeing another video on upscaling awhile ago (it was for SDXL) where the user was using a particular node where you can import a folder of images.....I don't remember off the top of my head but I'll make a note to check it out.
@@MonzonMedia I figured Comfy would be able to do batches but I'm not familiar enough with it to know and Forge has it built in so I use that.
I tend to save mine up, then when I have a few, do them all in one go over lunch or something. That was mainly with SD1.5 as I wanted higher resolutions but with SDXL and Flux I can get there straight away.
@@Elwaves2925 Totally hear ya. If I were to do batch runs then yes I'd do it on Forge since it's so much easier. I'm staying hopeful that the devs will come back to it soon. I really do like using Forge. Since it's a clone of A1111 it just feels natural to use since A1111 was the first local platform I learned.
great video! keep doing the same!
Thanks for watching! More videos like this are on the way!
Can I download your workflow? Thanks!
But I just showed how to build it from scratch! Best way to learn is to build it yourself. I added a link in the description to the workflow but make it a point to build it yourself otherwise you will never learn.
Whats your preference, Invoke Ai or ComfyUI?
For images invokeai hands down, it's just easier and fits my workflow. Comfyui is great for all purpose things such as video and of course if I want a custom workflow I can create it in comyfui. I would always recommend invokeai to anyone that wants a good stable platform.
❤
👊😊Cheers bud! Always appreciate you!
Your voice has something kinda similar to Pedro Pascal's. I sometimes expect you to end your tutorials by "this is the way".
hehehe that's quite the compliment! I don't think I've been compared to Pedro Pascal before. 😁😊
it is better to show upscale of real photos. the old ones. for restoration purposes.
This workflow was just an intro to upscaling as part of this series. There is so much more we can cover. Thanks for the suggestion!
This simple workflow is designed specifically for AI-generated images and is not suitable for real photographs, especially older ones. For real-life photos, additional steps are required to reduce compression artifacts and denoise the image. You also need to incorporate extra stages for ControlNet; otherwise, the person in the photo may not retain their original appearance after upscaling. Additionally, if the source photo is too large, processing it as a single piece can lead to out-of-memory (OOM) errors. If you use tiling to avoid this, it may introduce seam artifacts across the image.
True but how can one learn how to do these things without the knowing and understanding the basics first? Which is the purpose of this video and part of the Comfyui series on this channel.