i like to do one thing different with wildcards. If I have a very detailed description that I like to reuse (as in, a specific character) I can create a wildcard with only one entry - all that description - and then just use the wildcard in my prompt. It will give me that character description every time because that's the only thing in the file. Saves typing and makes the prompt much simpler to read as you add the rest
Thank you for the video - please make a video on how to load prompts from a text file so that we don’t when to enter prompts - many options in stable diffusion available
I have some problems removing things in fooocues. For example i created a horse and it had a haflter one. I put it in inpaint/outpaint marked the halter and said: remove halter. But it replaces the halter with another halter
Hi there.. first of all, nice vid... but I was wondering would you consider making a vid explaining more of the basics of Focus... for example what are those things under model tab like "base model, refiner, lora".. what are those weights next to them, what happens if we set them to -2 or +2 ... and what are those things under developer debug mode that you keep changing :)... etc etc ... If you can explain those tabs in detail, I would really appreciate and I also believe it will serve a much wider audience...
Hey thanks for the video!! Is there another location to download your wildcards? The website seems to be broken?? Also don't give up on making videos, your videos are fantastic!
hmmm quite easy but something is wrong in my case it doestn work. tried a wildcard for different races (black hispanic etc.) in the txt file each entry is on a seperate line. But then in the Fooocus logs "race" is still there and not replaced.
My "log.html" file doesn't show the generated image, just the metadata. Nor does the "copy to clipboard" function work. What is the correct way to update the entire system?
Hi Peter. Fooocus should update when the batch file is launched. If not, be sure you are not using flags that disabled it. If you did not add any flags then it should auto update. If you have issues check out the tracker here and see if you can find the fix or open a new issue. github.com/lllyasviel/Fooocus/issues
Solved, sorta. I was viewing the file metadata in the "outputs" folder, where the relevant images do not appear and the "copy to clipboard" function does not work on my system. If I explicitly load log.html with a double click, the copy/paste parameters function works. But only on log.html files from 2023/12/24 onwards. Opening log.html files from earlier than that date do not provide a "copy to clipboard" function. No idea why. Not sure if I have disabled "auto update on launch" function. How can I determine if that flag has been set? Thanks for your help. Fooocus continues to delight, as do your tutorials.@@MindRenders.
just saw the comment below about the upscale. you can change the denoise strength of the upscale if you want to reduce the level of change (advanced / developer debug mode / Forced Overwrite of Denoising Strength of "Upscale")
awesome video - it succinctly describes how to use wildcards in the Fooocus GUI for AI generated images.. i wish there were more great tutorials like this.
Thank You no one make a video or explanation before, I agree with you when you upscale an image Fooocus gives you slighly different results, I am looking for a solution too.
after some experiment I realise > upscale (fast 2X) does not change anyting in the image. results are almost the same with Topaz Photo AI upscale or other free alternative ¨upscaly´´ almost identical. @@MindRenders
That is a extremely useful video. Thanks for sharing! Is possible to create a startup bat to store the model, settings and advance settings are you suggesting to avoid adding them each time? In addition to the wildcard notes, I really like those settings.
Not that I am aware of. You can however view the logs and click the copy to clipboard button on any image you like and paste it into the prompt input then click "Load Parameters" and go from there! Thanks for watching!
I just have watched your video explaining it. @@MindRenders . I hope you keep doing videos like this, they are super helpful. Most of the people only give superficial information that you can get after 30 minutes with the program.
Using relatively recent install on Windows 10. "Copy to clipboard" not working here. Using log.html, paste into prompt field returns previously copied irrelevant data. Same thing happens when copied from History Log, despite screen mssg reporting successful copy function. This same "return to previous gen parameters" function is available in InvokeIA and is reeeeely uselful. Any ideas? @@MindRenders
@@MindRenders You can type a template for prompting in line 366 of the en.json \Fooocus\language\). So if you create another copy of en.json, ofc with another name, you can choose a template with the startup file (use the flag --en1 jp by example) .
Hi, thank you for the helpful video! One question: Is there a way of adding multiple word prompts with commas as wildcards like: Red hair, long hair, curly hair Pink hair, short hair, straight hair So the red hair will always be long and curly and the pink hair always short and straight?
Thank you so much!! I am just wondering how you figure this out with wild card? I mean how did you know if this option even exist? You are genius! Can I ask a question? is it possible this ai can make a picture based on my whole body including face, body shape and leg, then keep changing cloth design, which are separately given to AI by uploading picture of cloth itself or picture of someone wear that. and It also shows different body posing with same cloth on. I know I am asking too much :) I tried many ways already for a week, and no luck. So, basically I upload one design, then AI shows that on shirts while person is posing different ways in random. Possible?
Switch your desktop resolution to 1080p before recording a video. It's bad manners to use higher resolutions for youtube for educational purposes. The most users use simple Full-Hd monitors. Keep this in mind
i like to do one thing different with wildcards. If I have a very detailed description that I like to reuse (as in, a specific character) I can create a wildcard with only one entry - all that description - and then just use the wildcard in my prompt. It will give me that character description every time because that's the only thing in the file. Saves typing and makes the prompt much simpler to read as you add the rest
Clever. I thought about this AFTER I made the video too haha.
Promo'SM 💔
Thank you for the video - please make a video on how to load prompts from a text file so that we don’t when to enter prompts - many options in stable diffusion available
Nice, never knew about this feature, real fun to play with.
I have some problems removing things in fooocues. For example i created a horse and it had a haflter one. I put it in inpaint/outpaint marked the halter and said: remove halter. But it replaces the halter with another halter
Can you please do a video explaining what LoRA is and how it works... many thanks, great videos.
I'd be happy to! Give me a couple days to craft something up :)
Superb tutorial. Huge functionality. Subbed immediately.
Hi there.. first of all, nice vid... but I was wondering would you consider making a vid explaining more of the basics of Focus... for example what are those things under model tab like "base model, refiner, lora".. what are those weights next to them, what happens if we set them to -2 or +2 ... and what are those things under developer debug mode that you keep changing :)... etc etc ... If you can explain those tabs in detail, I would really appreciate and I also believe it will serve a much wider audience...
Thanks for the tips, I can't wait to play around with this.
Hey thanks for the video!! Is there another location to download your wildcards? The website seems to be broken?? Also don't give up on making videos, your videos are fantastic!
Your website is interesting too. Good job.
Thank you!
I prove to create a single wildcard of 200 detailed prompt but it doesn't work. Why? Your help will be appreciated.
Thanks for the video. As a side note, how did you enable the output log in HTML?
It’s in the fooocus folder where the images are saved I believe. There’s an html file there.
hmmm quite easy but something is wrong in my case it doestn work. tried a wildcard for different races (black hispanic etc.) in the txt file each entry is on a seperate line. But then in the Fooocus logs "race" is still there and not replaced.
Wow ... I had no clue. Ironically I couldn't find anything useful on how to use the wildcards
My "log.html" file doesn't show the generated image, just the metadata. Nor does the "copy to clipboard" function work. What is the correct way to update the entire system?
Hi Peter. Fooocus should update when the batch file is launched. If not, be sure you are not using flags that disabled it. If you did not add any flags then it should auto update. If you have issues check out the tracker here and see if you can find the fix or open a new issue. github.com/lllyasviel/Fooocus/issues
Solved, sorta. I was viewing the file metadata in the "outputs" folder, where the relevant images do not appear and the "copy to clipboard" function does not work on my system.
If I explicitly load log.html with a double click, the copy/paste parameters function works. But only on log.html files from 2023/12/24 onwards. Opening log.html files from earlier than that date do not provide a "copy to clipboard" function. No idea why.
Not sure if I have disabled "auto update on launch" function. How can I determine if that flag has been set?
Thanks for your help. Fooocus continues to delight, as do your tutorials.@@MindRenders.
how did you ask to chatgpt for shape list.What was your question
Just as follows: "list 50 different types of abstract shapes. dont use numbers, bullet points or other formatting. simply list them one on each line."
thank you very much@@MindRenders
just saw the comment below about the upscale. you can change the denoise strength of the upscale if you want to reduce the level of change (advanced / developer debug mode / Forced Overwrite of Denoising Strength of "Upscale")
awesome video - it succinctly describes how to use wildcards in the Fooocus GUI for AI generated images.. i wish there were more great tutorials like this.
Thank you! This made my day :)
Thank You no one make a video or explanation before, I agree with you when you upscale an image Fooocus gives you slighly different results, I am looking for a solution too.
I ran some tests today doing upcales in fooocus. It wasn’t as bad as I thought!
after some experiment I realise > upscale (fast 2X) does not change anyting in the image. results are almost the same with Topaz Photo AI upscale or other free alternative ¨upscaly´´ almost identical. @@MindRenders
great video, used to use this in a1111. I wonder if there is a way of making the selection of the wild cards incremental rather than random.
Thank you! Not that I’m aware of.
That is a extremely useful video. Thanks for sharing!
Is possible to create a startup bat to store the model, settings and advance settings are you suggesting to avoid adding them each time? In addition to the wildcard notes, I really like those settings.
Not that I am aware of. You can however view the logs and click the copy to clipboard button on any image you like and paste it into the prompt input then click "Load Parameters" and go from there! Thanks for watching!
I just have watched your video explaining it. @@MindRenders . I hope you keep doing videos like this, they are super helpful. Most of the people only give superficial information that you can get after 30 minutes with the program.
Using relatively recent install on Windows 10.
"Copy to clipboard" not working here. Using log.html, paste into prompt field returns previously copied irrelevant data. Same thing happens when copied from History Log, despite screen mssg reporting successful copy function.
This same "return to previous gen parameters" function is available in InvokeIA and is reeeeely uselful. Any ideas? @@MindRenders
@@MindRenders You can type a template for prompting in line 366 of the en.json \Fooocus\language\).
So if you create another copy of en.json, ofc with another name, you can choose a template with the startup file (use the flag --en1 jp by example) .
Hi, thank you for the helpful video! One question:
Is there a way of adding multiple word prompts with commas as wildcards like:
Red hair, long hair, curly hair
Pink hair, short hair, straight hair
So the red hair will always be long and curly and the pink hair always short and straight?
Thank you so much!! I am just wondering how you figure this out with wild card? I mean how did you know if this option even exist? You are genius!
Can I ask a question? is it possible this ai can make a picture based on my whole body including face, body shape and leg, then keep changing cloth design, which are separately given to AI by uploading picture of cloth itself or picture of someone wear that. and It also shows different body posing with same cloth on. I know I am asking too much :)
I tried many ways already for a week, and no luck. So, basically I upload one design, then AI shows that on shirts while person is posing different ways in random. Possible?
This is great! Thank you for the great video! I use wildcards quite a bit and am really glad that Fooocus has wildcard support!
Giga pixels 99$?
I think so. Try upscayl, it’s free and open source. Upscayl.org.
Is there a way to create consistent characters and also set poses?
nice
Is it possible to add text somehow to these images
You can try with AI but it doesn't usually end up well. I use PhotoPea for adding text after the images have been generated.
Switch your desktop resolution to 1080p before recording a video. It's bad manners to use higher resolutions for youtube for educational purposes. The most users use simple Full-Hd monitors. Keep this in mind
no. how's that for manners?