Download the workflow from Discord discord.gg/gggpkVgBf3 look for the channel pixaroma-workflows If you have any questions you can post them on discord or Pixaroma Community Group facebook.com/groups/pixaromacrafts/ ComfyUI Tutorial Series Playlist: ruclips.net/p/PL-pohOSaL8P9kLZP8tQ1K1QWdZEgwiBM0
Just a thank you to show my appreciation for your tutorial style and work flows that you provide. They really help in learning this wonderful image creator. I hope you continue posting more Comfy Tutorials.
15 000+ views but 700+ likes only. Common people, we are lucky to have this kind of tutorial for free. At least like and comment. Thank you so much, Pixaroma! You are a legend😇!
Just a token of my appreciation for teaching us about and sharing that AMAZING txt2img styles workflow. It would be fascinating to have something similar for FLUX…
@@pixaroma I was able to combine the flux version with style selection with parts of a workflow from Matteo of Latent Vision to add information such as seed etc. as well as the concatenated text prompt as a footer to the saved image. I also added the ultimate SD upscaler to my workflow(graph). If this sounds interesting how can I share this with you?
Nice update, thanks. For macOS noobs, we use forward slash for the path. If you can't get the path right, open up terminal and drop the style.csv file in terminal window from it's location in your ComfyUI folder, it will display the correct path you need.
Been doing basics and this is the first one that had me spinning lol. I currently don't need this workflow, but I'm pretty sure I will need to come back to this once the need arrives. Thank you so much! update after completing the video, aaaaah seeing it in action, I now get it, time to rewatch the whole thing lol!
This is probably (so far) the single most valuable thing I've learned (besides the basics). Not only can I make my own styles, I can use the ones you provided. TYSM! it took me a bit, but i got it working. I had issues realizing that text concat becomes the middle man between easy pos/neg and the original CLIP encode. IDKY. I watched it multiple times before I saw that. But I remembered a vid you did that had a segment focused on concat (not this one), which is the first time I really realized what it means, where it goes, and how it works, and I was able to figure this out. Happiness!
if you are not able to find Badge option is the managar, then its because u r on the lastest version of confyui where the badge is enabled by default. one thing to note here is tht instead of fox badge, the comfyui nodes are labeled #1, #2, #3....and so on
I am not kidding, I watched many ComfyUI tutorials and always got stuck somewhere (red nodes everywhere) and gave up for months and returned to A1111 - until your series.
Great work. All i added cause I am curious was to ad Show Text node s after each of the Text Concatenate nodes so I could read what the Syles Output said
I will do inpainting probably next month, I try to make the video in order, basics first then more editing and modifications, still need to cover some upscaling also, a lot to do just so little time to do all that, all the research, testing and video creation takes time, so i will get there eventually
very nice tutorial and fun to follow, I was able to find a text preview node fromComfyUI_Custom_Nodes_AlekPet author AlekPet from the manager, there'a a node in there called text preview and it does exactly that. EDIT: Also the badge name option is moved to the main comfyUI setting instead of the manager)
hello! thank you for this course, you explained everything perfectly! But i have a question, in an image2image workflow, which value should i modify to obtain an image 100% faithful to the original (as shape and contours)?
you can not do with default settings, you can have a little control if you use control net like canny and depth to maintain similar composition and shapes, but is not 100% the same, AI likes to do things random not the same :)
Hey, thanks for the styles . But when i use the styles with sdxl models like juggernaut or dreamshaper, they were perfect , but when i use them with flux nf4 v2 , 50% of the time i get realistic images(when i wanted painting style). Do the styles need an update for flux. if yes can you help with that?. thanks.
flux is not so good with art styles, it doesnt know so many art styles like sdxl, you can try to reduce the flux guidance to 2 or 2.5 for paintings but the problem is with flux, it just doesn't know how to do some art styles, hope they will fix that in next models
@pixaroma First off, great tutorials I'm still learning what is possible with comyui MariaScott Nodes has a node called Display Info Text v1 /I or similar, this one displays the text if plugged straight into the prompt selector, or into the end of the string output from concat node. I've left it in the work flow between concate and Clip text encode nodes I'm sure there are others but this works for me
there is a node called "show text" that works like "preview image" for showin text. I am not expert, and I don't know if I install something to have that node or it is a confyui default node but that will show the concatenated text
10:35 when I use the Prompt Multiple Styles Selector, the drop down menu doesn't appear. It works again when I use the Load Style CSV Node, Did anyone else face this same issue. Please help me out.
apart from this workflow i also liked that 0:03 transition although i have made a custom preset like this in resolve but like to know how did you applied this transition??
if u are looking for a Node to see the 'preview text ' like the 'preview image', you can install the custom node "ComfyUI-Easy-Use" and use the node called "Show Any" this is what pixaroma uses in his future tutorials. i came back to this tutorial just to add this comment.
I've tried your stylesheet and used it for a long time, mostly with photography styles for realistic renderings. I was wondering if it was possible to reduce the excessive bokeh rendering that Flux produces for most photographic images. Is there any way to change this setting?
As long as the focus is on something the background is blurred , unless it is a landscape and is using a wide lens. You can try a lora to reduce that civitai.com/models/675581/anti-blur-flux-lora
When I try to use the prompt multiple styles selector node, the message "none" appears instead of styles. What should I do? I couldn't find a solution.
can you post on discord probably you missed a step, post some screenshots there on the comfyui channel, the link to discord is on the youtube header where the links are located
The styles Are just prompts organized, is not like a Lora that let you change things,, so if is high denoise would look different, you can combine maybe with control net to keep composition, so you can use higher denoise, but stil Lora have higher effect in my opinion combined with right prompt
Episode 8 from next week would be about flux, so far is my favorite model, you need a good video card though to run the best one but is also one version that i got it to run on a 6gb of vram and 32 system ram
Is still new and need more time for lora to appear, each lora must be created from the same model base, so sdxl lora works only with sdxl, flux works only with flux lora, i think it needs more time for lora and control net to appear, i think i only saw one flux lor so far.. also probably only schnell model can be used, the dev has different license
Lost me here, this made no sense. too many duplicate text prompts and then you have whatever was written in them hidden when they are converted.. Idk why you wouldnt want to always have the prompt available to be changed. but maybe this workflow is not for me. The loras make a lot more sense for adding styles.
This are a collection of prompts that work with any model and is good for exploring a lot of styles, Lora is better for one targeted style for a specific bade model. Before I had chatgpt I used them a lot, now i just do complex prompts for flux with chatgpt exactly how I want them.
@@wervil yes, flux can handle really long prompts, that why I use chatgpt to get long one like sometimes a few sentences describing accurate. For sdxl and other need shorter and can not do so complex, like in sdxl I can not say a cat with blue hat, a red dress, sitting in a yellow car, because it will get all mixed up, but in flux I can do that and combine a lot of details, so give flux details about anything, subject, environment, how is dress, what color etc
thank you, I found since then a few nodes some with show text, I will see which one I will use, since in next episode I need to work more with text. But probably show any since I already have Easy Use node :)
Download the workflow from Discord
discord.gg/gggpkVgBf3
look for the channel pixaroma-workflows
If you have any questions you can post them on discord or Pixaroma Community Group facebook.com/groups/pixaromacrafts/
ComfyUI Tutorial Series Playlist: ruclips.net/p/PL-pohOSaL8P9kLZP8tQ1K1QWdZEgwiBM0
Your videos are a lot better than all of Comfyui courses in Udemy.
Just a thank you to show my appreciation for your tutorial style and work flows that you provide. They really help in learning this wonderful image creator. I hope you continue posting more Comfy Tutorials.
Thank you so much for support, yes I am working on more videos each week :)
@@pixaroma Thank you.
Thank you also from my part.
The undisputed king of comfyui tutorials 🙇♂️ thank you!
15 000+ views but 700+ likes only. Common people, we are lucky to have this kind of tutorial for free. At least like and comment.
Thank you so much, Pixaroma! You are a legend😇!
thanks 🙂
Just a small thank you to show my support. I really appreciate all your tutorial videos. Please keep up the great work!
Thank you for support , will do ☺️
just completed ep7.
every video that I watch, I m learning something new
Glad it helps ☺️
Extremely valuable and concise information. Thank you for your work.
Just a token of my appreciation for teaching us about and sharing that AMAZING txt2img styles workflow.
It would be fascinating to have something similar for FLUX…
Thank you so much, you can use it with flux check the episode 8 and the workflows on the discord ☺️
@@pixaroma I was able to combine the flux version with style selection with parts of a workflow from Matteo of Latent Vision to add information such as seed etc. as well as the concatenated text prompt as a footer to the saved image. I also added the ultimate SD upscaler to my workflow(graph). If this sounds interesting how can I share this with you?
Thanks, the easiest way is probably on discord, there you can also attach the json workflow :) discord.gg/a8ZM7Qtsqq
❤❤
Thank you for the detailed walk through. Your patient and detailed explanations are very much appreciated. I look forward to more.
Nice update, thanks. For macOS noobs, we use forward slash for the path. If you can't get the path right, open up terminal and drop the style.csv file in terminal window from it's location in your ComfyUI folder, it will display the correct path you need.
Thank you so much for this video. Extremely valuable info and techniques. And thanks for your work on the styles file.
Been doing basics and this is the first one that had me spinning lol. I currently don't need this workflow, but I'm pretty sure I will need to come back to this once the need arrives. Thank you so much!
update after completing the video, aaaaah seeing it in action, I now get it, time to rewatch the whole thing lol!
Downloaded the workflow from DIscord. Thank you for all the resources.
this tutorial series is amazing! thank you so much.
Glad you enjoy it 🙂
This is probably (so far) the single most valuable thing I've learned (besides the basics). Not only can I make my own styles, I can use the ones you provided. TYSM!
it took me a bit, but i got it working. I had issues realizing that text concat becomes the middle man between easy pos/neg and the original CLIP encode. IDKY. I watched it multiple times before I saw that. But I remembered a vid you did that had a segment focused on concat (not this one), which is the first time I really realized what it means, where it goes, and how it works, and I was able to figure this out. Happiness!
Thank you, You can also try the styles node from ep16 ☺️
Thank you for your amazing comfyui tutorial series, truly appreciate your effort in sharing this knowledge!
loving the series! tthe discord downloads are an amazing extension
Very well done, slow and clear explanations
There should be a extra 0 or 2 after the followers. Absolut amazing work you are doing. Even for a non-english guy it's easy to follow, an learning 😀
man you're the best! Please keep going with these amazing tut!
if you are not able to find Badge option is the managar, then its because u r on the lastest version of confyui where the badge is enabled by default.
one thing to note here is tht instead of fox badge, the comfyui nodes are labeled #1, #2, #3....and so on
There are in settings now, search badge and you have more options, i have screenshots and instructions on discord on comfyui-faq channel
@@pixaroma Thaks for the update. u r the best😇
Thank you. I learned a lot from this great video. I look forward to the next video.
I am not kidding, I watched many ComfyUI tutorials and always got stuck somewhere (red nodes everywhere) and gave up for months and returned to A1111 - until your series.
having learned some python in the past really helps understanding many concepts of this.
Great video!! Thank you very much...
Super Useful!!
So many tools, and it only gets better...
take me some time to try . finally i manage to make it work and combine with image control net.
yes ! thank you so much
this video is absolute gold
:D thanks you
Wooow this is just Amazing !
Great work. All i added cause I am curious was to ad Show Text node s after each of the Text Concatenate nodes so I could read what the Syles Output said
yes I use Show Any node to preview also :)
I finally managed to make a workflow all thanks to your videos! Can you please also make video on how to add face detailer in workflow and inpainting?
I will do inpainting probably next month, I try to make the video in order, basics first then more editing and modifications, still need to cover some upscaling also, a lot to do just so little time to do all that, all the research, testing and video creation takes time, so i will get there eventually
very nice tutorial and fun to follow, I was able to find a text preview node fromComfyUI_Custom_Nodes_AlekPet author AlekPet from the manager, there'a a node in there called text preview and it does exactly that.
EDIT: Also the badge name option is moved to the main comfyUI setting instead of the manager)
Hi! To view text there are nodes
Show Text from ComfyUI-Custom-Scripts
Show Text from comfyui-mixlab-nodes
Textbox from ComfyUI-Chibi-Nodes
Thank you ☺️
Genial !🤩
Thanks! you are great but a little bit complicated in this initial part of the course ;)
many thanx man
perfect
thx
hello! thank you for this course, you explained everything perfectly! But i have a question, in an image2image workflow, which value should i modify to obtain an image 100% faithful to the original (as shape and contours)?
you can not do with default settings, you can have a little control if you use control net like canny and depth to maintain similar composition and shapes, but is not 100% the same, AI likes to do things random not the same :)
@@pixaroma okay I'll try control net 😁 thank youuu
Hey, thanks for the styles . But when i use the styles with sdxl models like juggernaut or dreamshaper, they were perfect , but when i use them with flux nf4 v2 , 50% of the time i get realistic images(when i wanted painting style). Do the styles need an update for flux. if yes can you help with that?. thanks.
flux is not so good with art styles, it doesnt know so many art styles like sdxl, you can try to reduce the flux guidance to 2 or 2.5 for paintings but the problem is with flux, it just doesn't know how to do some art styles, hope they will fix that in next models
@@pixaroma oh Thanks. It's just that the context in the images with flux is good.hope they update flux.
@pixaroma First off, great tutorials I'm still learning what is possible with comyui
MariaScott Nodes has a node called Display Info Text v1 /I or similar, this one displays the text if plugged straight into the prompt selector, or into the end of the string output from concat node.
I've left it in the work flow between concate and Clip text encode nodes
I'm sure there are others but this works for me
Thank you :) I will check it out
there is a node called "show text" that works like "preview image" for showin text. I am not expert, and I don't know if I install something to have that node or it is a confyui default node but that will show the concatenated text
I got a node that has Show Any from Easy-Use custom node, that I use it in future episode. Thanks :)
1:36 I can't find "badge nickname" I'm in the latest version of comfyui rn
watch this from minute 3 58 I explain there ruclips.net/video/VpW5r06Xb5g/видео.htmlsi=1N-_OOM90FHsZYsf&t=238
@pixaroma thank youuu
10:35 when I use the Prompt Multiple Styles Selector, the drop down menu doesn't appear. It works again when I use the Load Style CSV Node, Did anyone else face this same issue. Please help me out.
I just tested today a workflow with multiple styles and it worked for me
Maybe try the one from episode 15 if ep07 doesn't work
apart from this workflow i also liked that 0:03 transition although i have made a custom preset like this in resolve but like to know how did you applied this transition??
In capcut desktop pro, on the transitions tab there is one called Pull Away that create that effect between 2 images or videos
if u are looking for a Node to see the 'preview text ' like the 'preview image', you can install the custom node "ComfyUI-Easy-Use" and use the node called "Show Any"
this is what pixaroma uses in his future tutorials.
i came back to this tutorial just to add this comment.
Yes that is the one i use in my other episodes:)
I've tried your stylesheet and used it for a long time, mostly with photography styles for realistic renderings. I was wondering if it was possible to reduce the excessive bokeh rendering that Flux produces for most photographic images. Is there any way to change this setting?
As long as the focus is on something the background is blurred , unless it is a landscape and is using a wide lens. You can try a lora to reduce that civitai.com/models/675581/anti-blur-flux-lora
Hello, very useful videos! May I ask how should I set the style file path if I'm using a MacBook?
I don't have a mac so not sure but must be a way to take the path
text concat and easy positive... can i suggest to use wd14 tagger?
I will look into it, thanks
@@pixaroma i mean string function by pythongoossss.. sorry :)
ComfyUI Manager, there is no "badge: none" section what should I do
Check the beginning of episode 18 i talk there about new interface is now on settings
When I try to use the prompt multiple styles selector node, the message "none" appears instead of styles. What should I do? I couldn't find a solution.
can you post on discord probably you missed a step, post some screenshots there on the comfyui channel, the link to discord is on the youtube header where the links are located
Can i change image to image with this workflow
The styles Are just prompts organized, is not like a Lora that let you change things,, so if is high denoise would look different, you can combine maybe with control net to keep composition, so you can use higher denoise, but stil Lora have higher effect in my opinion combined with right prompt
@pixaroma thanks for replying. Very helpful for a newbie like me to learn
Hi. What can you tell us about Flux? For some reason it's almost worshipped on Civitai.
Episode 8 from next week would be about flux, so far is my favorite model, you need a good video card though to run the best one but is also one version that i got it to run on a 6gb of vram and 32 system ram
@@pixaroma I have a 3050 8gb. I'll be looking forward to your video. Because 50 buzz-points in Civitai for one photo for FLUX work is too much.
Will it be used for character creation? How does it work with Lora's? Or is it mainly for designers and ad men?
Is still new and need more time for lora to appear, each lora must be created from the same model base, so sdxl lora works only with sdxl, flux works only with flux lora, i think it needs more time for lora and control net to appear, i think i only saw one flux lor so far.. also probably only schnell model can be used, the dev has different license
Preview text in output node: Custom_Nodes_AlekPet.
thanks, I am using now Show Any from ComfyUI Easy Use
Sorry, i don't have Badge line in manager (
They changed the interface a few days ago, i am not at pc this weekend to check it, but must be somewhere in the settings
@@pixaroma don't worry! You so amazing! Thank you!
It's in the main settings now 🙂
SHOW TEXT from custom scripts can show you texts
I'm sure it's been mentioned a million times but too lazy to search - you can use text to console to view the string in the console.
It has only loaded 218 styles. However, the file has 309 styles.
That is strange i didn't count it after, maybe the node is doing something, i have to check. Thanks
@@pixaroma i didn't count them either but i saw on the cmd terminal only 218 styles loaded
@@tranthai762 I think that refers to the nodes, that are part of the was node suite, it says nodes not styles :)
ShowText|pysssss node shows text
Thanks I use ShowAny node in future videos
Lost me here, this made no sense. too many duplicate text prompts and then you have whatever was written in them hidden when they are converted.. Idk why you wouldnt want to always have the prompt available to be changed. but maybe this workflow is not for me. The loras make a lot more sense for adding styles.
This are a collection of prompts that work with any model and is good for exploring a lot of styles, Lora is better for one targeted style for a specific bade model. Before I had chatgpt I used them a lot, now i just do complex prompts for flux with chatgpt exactly how I want them.
@@pixaroma interesting, so Flux can handle more complex prompt writing? Obviously I'm somewhat a newbie so i appreciate your reply.
@@wervil yes, flux can handle really long prompts, that why I use chatgpt to get long one like sometimes a few sentences describing accurate. For sdxl and other need shorter and can not do so complex, like in sdxl I can not say a cat with blue hat, a red dress, sitting in a yellow car, because it will get all mixed up, but in flux I can do that and combine a lot of details, so give flux details about anything, subject, environment, how is dress, what color etc
@pixaroma To preview text/strings/anything you can use the custom node "Show Any" (EasyUse > Logic > Show Any)
thank you, I found since then a few nodes some with show text, I will see which one I will use, since in next episode I need to work more with text. But probably show any since I already have Easy Use node :)
I don't get additional options on a right click...
Not sure on what you right click since are so many things in comfy UI, post on discord a screenshot