@@sergiogonzalez2611 I don't even know what I did but I checked every option to squeeze a little bit more performance out 🤷🏼♂️ (for example using --lowvram as a start argument). After that it worked without lora. Also I closed a lot of apps running in the background so before even running dreambooth I had only 0.5 GB of vram being used.
I can explain what to do clearly too; "put yer head under me skirt and stick yer tongue out, and feel around for something warm wet and hairy. between me legs... You feel something crawling on your cheeks its just me crabs". That is my prompt for a viral video by Dreambooth and Unstable Illusion
CGMatter to the rescue! I every very glad you are tackling this. I've been on this Stable Diffusion boat for 4 months now and navigating around has been challenging with stuff changing every week. I can hardly wait to merge Blender and SD into making some great things. I still have your Geometry Nodes on my queue too.
Aah like things to the point. But to get the the point... the fella in this video never revealed where the dreambooth is, nor did he tell us how much it is an hour to go dream in the booth. So where can thee go to have a dream of your choosing, in a booth? Aah still don't inderstand all this modern newfandangled words.But I want to have a go in that Dream Booth. Can you choose what you dream about? How much does it cost?
Can anyone help me please? I have done everything like in video. I have tried it many times. But everytime it writes me same error: Exception training model: ''NoneType' object is not subscriptable'. What do do?
Is it necessary to have the images in 512*512? If I want to train a model, are the images also in 512*512? What happens if you use the actual size photos?"
How long does this process take? At my place it goes a long time and I only have 4/800 steps. I have 24Gb RAM. Is it possible to train a patria model, such as 3 hours each day?
I encountered an issue that says 'Exception training model:~/Download/*image file* not in list', and upon testing, I found that this issue occurs when the 'Cache Latents' option is turned on. Can you help me understand why this is happening?
Many thanks for this tutorial. Really helped me to understand which 80% of settings are most important just to get started. How do I create a LORA instead? Is it as simple as ticking the 'use LORA button'? I always get errors when checking that box.
Aah just thought of a prompt for a video in dreambook horse-stable disillusion; "The leaders of all nations dressinf in short pants, bent over a bench with Bad Grandma giving them all a whipping. using a barbed wire wrapped cricket bat to shve up their butts" That would be a lovely video. Aaah know aah'm a naughty bad Grandma Aaah am. Aaah never went to school you know? I had to work so long every day that i used to get back from work just as i was getting ready to leave.
You can also use 768^2 images with the 1.5 model but it requires more vram because the higher image resolution increases the number of pixels by more than double. 262k vs 589k pixel
Getting this error: Exception training model: 'Repo id must use alphanumeric chars or '-', '_', '.', '--' and '..' are forbidden, '-' and '.' cannot start or end the name, max length is 96: 'C:\Users\Creed\stable-diffusion-webui\models\dreambooth\SlaveLeiaBikini\working\tokenizer'.'. Do you know what's wrong?
I can't get it to work. When I click "train" I get the message "Please configure some concepts" and nothing happens. I have the impression that I need to enter more information into the "concepts" tab, but I don't understand all those options. Also under advanced settings where you had "8 bit Adam", "fp16" and "xformers", I didn't have all those options. Closest I could get was "8 bit AdamW", "fp16" and "default" I also tried using the 1.5 pruned.ckpt, but that gave me the error: "Missing model directory, removing model: models/dreambooth/[my modelname]/working/unet". Going to 1.4 fixed this. So I don't know what to do.
I got the same message. However I successfully created 3 models before this last attempt and they worked great. The last one that worked was earlier this week which was less than 6 days ago, so it can't be an update.
I successfully created 3 models from this tutorial, thank you. Unfortunately it all of a sudden stopped working. I'm trying to create a new one and whenever I click the Train button, I get "Please configure some concepts" in the right column under "Output". Did something change recently? I followed this video precisely.
The merge is surprisingly easy. Have 2 models you want to merge and play with the weights for either one and let it run. Just tried it recently and works
I get the following error when hitting "create model:" Repo id must use alphanumeric chars or '-', '_', '.', '--' and '..' are forbidden, '-' and '.' cannot start or end the name, max length is 96: (And then my path to the source checkpoint) I've made sure the overall length is under 96, so I think it's throwing an error because it's referencing the C:\\ drive and colons are not allowed. I have no idea how to get around this though because it's stored locally and I need to reference the hard drive!
I got stuck after I clicked "Create Model", There are some error message like "ERROR: Exception in ASGI application", at Output it stopped at 18% ETA: 42s, any ideas?
Anyone else getting this errror? RuntimeError: CUDA error: an illegal memory access was encountered CUDA kernel errors might be asynchronously reported at some other API call,so the stacktrace below might be incorrect. For debugging consider passing CUDA_LAUNCH_BLOCKING=1.
Very superficial tutorial, and don't tell me that "it was for the sake of simplicity/easiness" because you also skipped fundamental infos on most of the options you covered.
I have issue about Dreambooth. First: I can't create model when I choose some of checkpoints. But some of can create model. Second: In memory attention tab, 'default' is the only option. there is no 'x-formers' :/. I'm using MacBook Pro M1 Pro.. Is there anyone had this issue too? PS. great tutorial btw.
Hello everyone. Has anyone had this issue? The following appears when I select train. When trainingException training model: ''/Users/miguelburlecoria/Desktop/Dreambooth/4.png' is not in list'.
I love this series about AIs ..but my pc can't run this softwares so please can u make a tut how to setup image generator on google cloud(cuz it's free at sm point) ..or something.. Thanks.
i get this error at 4:33 Exception training model: 'Repo id must use alphanumeric chars or '-', '_', '.', '--' and '..' are forbidden, '-' and '.' cannot start or end the name, max length is 96: 'C:\Users\21694\OneDrive\Desktop\Image gen\Web ui\stable-diffusion-webui\models\dreambooth\CCmatthias\working\tokenizer'.'.
how to install it properly, im using google colab notebook automatic1111 stable diffusion, i tried to install it using Extensions/available/github/restart the whole program, the file is already on my sd/sd-webui/extensions, but it's not appearing on the UI
Is every checkpoint another 7 gig or will be 14 gb and if the the first 7/14 gb important after I create another ? if I can delete it and use the newer one
I going to say yes but test it. Just copy the checkpoint to another folder, remove the original and see if your 2nd or 3rd one still works. If not you can replace your 1st one and if so you can delete it.
This is one o them things aah do't understand wha't me great great great great grandson is always on about. I dont understand all of this stuff, dont know what young people are talkng about these days. Aah do't see any connection between horse stables and nuclear fusion. I dont think nuclear diffusion sounds stable or safe at all. People keep talking about sme GPT formula 3 racing car all the time too saying it can talk. Cars cant blooming talk!
![I Replaced ALL my ADOBE APPS with these [free or cheaper] Alternatives!](http://i.ytimg.com/vi/5EfqHg49kMk/mqdefault.jpg)
you literally have THEE most helpful dreambooth tutorial on youtube, I can't thank you enough!
your simple breakdowns are refreshing
any way to get the ckpt to a 2gb'ish size, from within the A1111 dreambooth interface?
@@ronnykhalil yes, when he say you can check half size, it will output a 2gb size file
Step one is "click the Dreambooth tab," but he doesn't explain how to get a Dreambooth tab. Absolutely useless.
You used 13.2 GB of VRAM. That's an important requisite.
I got it running on a 12gb 3060, but I had to do a little bit of optimizing
What kinda of gpu does he have....
@@LauLauHip what do you do? i have a 3080 12 gigas de vram
@@sergiogonzalez2611 I don't even know what I did but I checked every option to squeeze a little bit more performance out 🤷🏼♂️ (for example using --lowvram as a start argument). After that it worked without lora. Also I closed a lot of apps running in the background so before even running dreambooth I had only 0.5 GB of vram being used.
3080 10gb worked just fine
This was surprisingly clear and easy to follow! THANKS!
I can explain what to do clearly too; "put yer head under me skirt and stick yer tongue out, and feel around for something warm wet and hairy. between me legs... You feel something crawling on your cheeks its just me crabs". That is my prompt for a viral video by Dreambooth and Unstable Illusion
The thumbnail seems like a Holi special.
As always great tutorial.
amazing turorial, simple and direct. pleasee do a lora tutorial too
Thank you. I've tried other tutorials, but this one worked perfect.
worked flawlessly wit the new dreambooth update, thank you
CGMatter to the rescue! I every very glad you are tackling this. I've been on this Stable Diffusion boat for 4 months now and navigating around has been challenging with stuff changing every week. I can hardly wait to merge Blender and SD into making some great things. I still have your Geometry Nodes on my queue too.
More to-the-point tutorials like this please👌
Aah like things to the point. But to get the the point... the fella in this video never revealed where the dreambooth is, nor did he tell us how much it is an hour to go dream in the booth. So where can thee go to have a dream of your choosing, in a booth? Aah still don't inderstand all this modern newfandangled words.But I want to have a go in that Dream Booth. Can you choose what you dream about? How much does it cost?
Nice quick tutorial, but your GPU VRAM limits a lot of us from doing the same settings. Which card are you running?
Dude you'd be right at home with ComfyUI, just like Blender Nodes. Hope to see a tut from you using ComfyUI and SDXL 1.0 someday.
straight to the point I love it! Keep it up
Can anyone help me please? I have done everything like in video. I have tried it many times. But everytime it writes me same error: Exception training model: ''NoneType' object is not subscriptable'.
What do do?
same here did u find a way?
you solved what no other tutorials solved for me. Thanks so much!
Great tutorial. Thanks!
super easy and on point.
Thanks for the tutorial man! Any tips for getting a better result? If i put more pictures does it helps?
Is it necessary to have the images in 512*512? If I want to train a model, are the images also in 512*512? What happens if you use the actual size photos?"
How long does this process take? At my place it goes a long time and I only have 4/800 steps. I have 24Gb RAM. Is it possible to train a patria model, such as 3 hours each day?
I encountered an issue that says 'Exception training model:~/Download/*image file* not in list', and upon testing, I found that this issue occurs when the 'Cache Latents' option is turned on. Can you help me understand why this is happening?
Training finished. Total lifetime steps: 0
Nice tutorial as always! Can you please do one on how to process video with Stable Diffusion?
Many thanks for this tutorial. Really helped me to understand which 80% of settings are most important just to get started.
How do I create a LORA instead? Is it as simple as ticking the 'use LORA button'? I always get errors when checking that box.
did someone got this error?
model_cfg: DreamboothConfig = Body(description="The config to save"),
NameError: name 'DreamboothConfig' is not defined
What would you say is better, using dream Booth or using the training method that comes with stable diffusion?
how do i install dreambooth?
Exception training model: 'No executable batch size found, reached zero.'.. what does this mean
I also got this one. Not sure what it mean.
same here stuck on it
@@tommysalami420 same...
Getting the same error.
im getting the same error, have you figured it out?
you're legend
I had some issues, but skipping xformers (and leaving default) did it for me.
Oh its Blender Guy!!!
Aah just thought of a prompt for a video in dreambook horse-stable disillusion; "The leaders of all nations dressinf in short pants, bent over a bench with Bad Grandma giving them all a whipping. using a barbed wire wrapped cricket bat to shve up their butts" That would be a lovely video. Aaah know aah'm a naughty bad Grandma Aaah am. Aaah never went to school you know? I had to work so long every day that i used to get back from work just as i was getting ready to leave.
I dont see the input tab in the middle, its changed some do you have a newer version of the video?
same issue here.
love the tutorial
Would it be better to make this small dataset a LoRA? Would you select SAVING > HALF LORA to do that?
orator
raise RuntimeError("No executable batch size found, reached zero.")
RuntimeError: No executable batch size found, reached zero.
Restored system models.
Duration: 00:00:57
if we are using 2.1 could we do 768x768 images/model? thanks for the tut fam, subscribed
You can also use 768^2 images with the 1.5 model but it requires more vram because the higher image resolution increases the number of pixels by more than double. 262k vs 589k pixel
Getting this error:
Exception training model: 'Repo id must use alphanumeric chars or '-', '_', '.', '--' and '..' are forbidden, '-' and '.' cannot start or end the name, max length is 96: 'C:\Users\Creed\stable-diffusion-webui\models\dreambooth\SlaveLeiaBikini\working\tokenizer'.'.
Do you know what's wrong?
me too
do u got the solution for this
Exception training model: 'No executable batch size found, reached zero.'....Iam getting error after everything Is done can anyone help me out?
im having the same issue, if you figured it out please help me.
@@kingpanda9846 did u find the solution?
why is it that i cant find xformer under memory attention?
Cheers m8, nice guide
I can't get it to work. When I click "train" I get the message "Please configure some concepts" and nothing happens. I have the impression that I need to enter more information into the "concepts" tab, but I don't understand all those options.
Also under advanced settings where you had "8 bit Adam", "fp16" and "xformers", I didn't have all those options.
Closest I could get was "8 bit AdamW", "fp16" and "default"
I also tried using the 1.5 pruned.ckpt, but that gave me the error: "Missing model directory, removing model: models/dreambooth/[my modelname]/working/unet". Going to 1.4 fixed this.
So I don't know what to do.
Same here
I got the same message. However I successfully created 3 models before this last attempt and they worked great. The last one that worked was earlier this week which was less than 6 days ago, so it can't be an update.
Moving the whole Stable Diffusion folder to C:\ fix the problem for me
Thank yoou
i finished modeling but there are no checkpoint, why did this happen
i have the same problem
Stuck on
"Initializing bucket counter!
Steps: 0%| | 0/300 [00:00
Can I share a video I uploaded where I'm trying to figure out how to make that video animation from photos?
ua-cam.com/users/shortsYo41aeWacEE
happy holi🙏
I was watching a 40min tutorial, then i cam here.
my dreambooth tab isnt appearing after installing...
This guy traped on ai.
Cool! Thanks.
Theres any tutorial to train my 8gb vram card to go to 24gb vram? :'D
Unfortunately, nope. You must rent GPU time on a collab notebook.
hey, i just doing it on collab here ua-cam.com/video/kCcXrmVk1F0/v-deo.html&ab_channel=MattWolfe
update, it worked
22min of traning
I successfully created 3 models from this tutorial, thank you. Unfortunately it all of a sudden stopped working. I'm trying to create a new one and whenever I click the Train button, I get "Please configure some concepts" in the right column under "Output". Did something change recently? I followed this video precisely.
you must presss "save settings" on your new model.
@@PiotrGarryWysocki I did that, no luck. I ended up reinstalling A1111 and it works now.
Please create a how-to on merge as well!
The merge is surprisingly easy. Have 2 models you want to merge and play with the weights for either one and let it run. Just tried it recently and works
but where and how do you load the images to work with???
ERROR - Exception training model: 'No executable batch size found, reached zero
nice and simple, ty!!! 4:13 TY for going over this part, I would just add that class Prompt should then say "a photo of person", right?
came baaaaack and subbed.
I get the following error when hitting "create model:"
Repo id must use alphanumeric chars or '-', '_', '.', '--' and '..' are forbidden, '-' and '.' cannot start or end the name, max length is 96: (And then my path to the source checkpoint)
I've made sure the overall length is under 96, so I think it's throwing an error because it's referencing the C:\\ drive and colons are not allowed. I have no idea how to get around this though because it's stored locally and I need to reference the hard drive!
No executable batch size found, reached zero :(
That's nice & all but I'm running on a 1070 with 8Gb Vram so ...
But what exactly is this DREAMBOOTH for? Images can be created on Txt2img and on img2img simply
For specific person, object, style, etc
@@1Know1tHurts Ahh,ok
I got stuck after I clicked "Create Model", There are some error message like "ERROR: Exception in ASGI application", at Output it stopped at 18% ETA: 42s, any ideas?
gold
Exception training model: "Module List'object has no attribute'2".
I get such an error, what should I do?
This for windows. For mac this tutorial won't work. Can you create a tutorial for M1 Mac?
can i train on bigger images then 512 ?
ModuleNotFoundError: import of xformers halted; None in sys.modules
anyone know how to fix this?
add --xformers flag to the init script
"Please check your dataset directories." - all I ever get. Data set is 512 and in a very straightforward directory.
same
Exception training model: 'No executable batch size found, reached zero
I keep getting an error that says exception training model not on list, what does this mean?
Anyone else getting this errror?
RuntimeError: CUDA error: an illegal memory access was encountered
CUDA kernel errors might be asynchronously reported at some other API call,so the stacktrace below might be incorrect.
For debugging consider passing CUDA_LAUNCH_BLOCKING=1.
probably too low vram
Very superficial tutorial, and don't tell me that "it was for the sake of simplicity/easiness" because you also skipped fundamental infos on most of the options you covered.
Noiceee
I have issue about Dreambooth.
First: I can't create model when I choose some of checkpoints. But some of can create model.
Second: In memory attention tab, 'default' is the only option. there is no 'x-formers' :/. I'm using MacBook Pro M1 Pro.. Is there anyone had this issue too? PS. great tutorial btw.
my a1111 missing xformer file so my train model always failed, do you know to you add? thanks
4:31 i get thiss error on the output tab RuntimeError: No executable batch size found, reached zero
Hello everyone. Has anyone had this issue? The following appears when I select train. When trainingException training model: ''/Users/miguelburlecoria/Desktop/Dreambooth/4.png' is not in list'.
How can I install dreambooth?? for some reason the dreambooth tab doesnt show up. Please tell me if you can! thank you
I love this series about AIs ..but my pc can't run this softwares so please can u make a tut how to setup image generator on google cloud(cuz it's free at sm point) ..or something..
Thanks.
for some reason i keep getting "Exception training model: ''NoneType' object is not subscriptable'."
When ı train its finishes in 10 seconds and says the folder name and is not a directory
YES!!!
i get this error at 4:33 Exception training model: 'Repo id must use alphanumeric chars or '-', '_', '.', '--' and '..' are forbidden, '-' and '.' cannot start or end the name, max length is 96: 'C:\Users\21694\OneDrive\Desktop\Image gen\Web ui\stable-diffusion-webui\models\dreambooth\CCmatthias\working\tokenizer'.'.
how to install it properly, im using google colab notebook automatic1111 stable diffusion, i tried to install it using Extensions/available/github/restart the whole program, the file is already on my sd/sd-webui/extensions, but it's not appearing on the UI
If I check "Use LORA", do I need to modify other settings?
Followed this tutorial EXACTLY - did not receive good results though 🤨
uhh, how do i install it then?
is there a way to do this in google collab?
I have 3090, why it show 6.9/10.5GB, is that mean only use 10GB?
Unable to find checkpoint file!
Duration: 00:00:00 i get this man.
i have the model
Is every checkpoint another 7 gig or will be 14 gb
and if the the first 7/14 gb important after I create another ? if I can delete it and use the newer one
I going to say yes but test it. Just copy the checkpoint to another folder, remove the original and see if your 2nd or 3rd one still works. If not you can replace your 1st one and if so you can delete it.
You can delete checkpoints you do not need/want when it's finished (from the model folder)
What GPU are you using?
why it doesn't work anymore?
This is one o them things aah do't understand wha't me great great great great grandson is always on about. I dont understand all of this stuff, dont know what young people are talkng about these days. Aah do't see any connection between horse stables and nuclear fusion. I dont think nuclear diffusion sounds stable or safe at all. People keep talking about sme GPT formula 3 racing car all the time too saying it can talk. Cars cant blooming talk!
duck 🦆
RuntimeError: No executable batch size found, reached zero --HELP
same problem :/
it doesnt work in my RTX 3060 12GB, if someone knows a workaround would be great.
does it work with a gtx 1650 4gb vram
Anyone else getting a 'nonetype' object is not subscriptable error?
Yes, me. Have you been able to solve it?