I re-did it, and made sure to caption my dataset, works great now!

BB0nd I re-did it, and made sure to caption my dataset, works great now!

F

Furkan Gözükara SECourses•12/23/23, 9:52 PM

great

1100Sachen80 Just did the patreon btw 🙂

F

Furkan Gözükara SECourses•12/23/23, 9:52 PM

i am glad you solved

Ttourist07 The same happened for me. Turn the samples creation off down to 0. Also I turned...

F

Furkan Gözükara SECourses•12/23/23, 9:53 PM

xformers reduces quality. off better

F

Furkan Gözükara SECourses•12/23/23, 9:53 PM

nice speed

DDeadMan1999 @Dr. Furkan Gözükara You should make a quick video on this: Nvidia has released ...

F

Furkan Gözükara SECourses•12/23/23, 10:07 PM

i agree. it starts using shared vram. but once you disabled it you will get OOM instead :d

F

Furkan Gözükara SECourses•12/23/23, 10:07 PM

in some cases it will be super useful

F

Furkan Gözükara SECourses•12/23/23, 10:07 PM

in some cases opposite

Vveldonlives 2 hardware performance questions: 1. I recently upgraded from an 2060 to a 3090...

F

Furkan Gözükara SECourses•12/23/23, 10:22 PM

remove --medvram or --lowvram if you had in auto1111

F

Furkan Gözükara SECourses•12/23/23, 10:22 PM

16 GB ram is low yes

F

Furkan Gözükara SECourses•12/23/23, 10:22 PM

try to get 32 gb

VVexmachina When using accelerate and sdxl_train.py from the commandline, instead of using t...

F

Furkan Gözükara SECourses•12/23/23, 10:22 PM

i always use gui

MMax Hello lads! It's always me, an error generator. I was following this video https...

F

Furkan Gözükara SECourses•12/23/23, 10:23 PM

this means it didnt find any imges in reg images folder

F

Furkan Gözükara SECourses•12/23/23, 10:23 PM

if you provided

Ggeno1237341 same issues

F

Furkan Gözükara SECourses•12/23/23, 10:23 PM

which issues

Vvp anyone know the outcome of triton on windows? I know it released recently and wa...

F

Furkan Gözükara SECourses•12/23/23, 10:24 PM

when i tested it didnt speed improve for me

MMax Sorry guys, I was trying to train a checkpoint using a custom model (GHArt/Reali...

F

Furkan Gözükara SECourses•12/23/23, 10:24 PM

in that folder no fp16 obviously

G

geno1237341•12/23/23, 10:24 PM

nono tahnk you, the only problem that i have is that i don't know how use another checkpoint

G

geno1237341•12/23/23, 10:25 PM

the code give me an erro r

G

geno1237341•12/23/23, 10:25 PM

error*

MMax Sorry guys, I was trying to train a checkpoint using a custom model (GHArt/Reali...

F

Furkan Gözükara SECourses•12/23/23, 10:25 PM

i would download its safetensors into kaggle and give its path

Ggeno1237341 nono tahnk you, the only problem that i have is that i don't know how use anothe...

F

Furkan Gözükara SECourses•12/23/23, 10:25 PM

where you want to use? can you explainmore

Ssainthell Hello... Might not be the right place to post this, but if I have 160 images of ...

F

Furkan Gözükara SECourses•12/23/23, 10:25 PM

160 is quite a lot. you can train up to 100 and save checkpoints every 20 and see

FFurkan Gözükara SECourses i would download its safetensors into kaggle and give its path

G

geno1237341•12/23/23, 10:25 PM

same issues of @doggo

FFurkan Gözükara SECourses i would download its safetensors into kaggle and give its path

G

geno1237341•12/23/23, 10:26 PM

ohh, okkk

EEyeSpyBekoAI Hey there community. A quick random question for you. Besides for perhaps train...

F

Furkan Gözükara SECourses•12/23/23, 10:26 PM

for speed

F

Furkan Gözükara SECourses•12/23/23, 10:26 PM

if you need speed A6000 best

F

Furkan Gözükara SECourses•12/23/23, 10:26 PM

and higher batch size

G

geno1237341•12/23/23, 10:26 PM

thank you very much master

FFurkan Gözükara SECourses i would download its safetensors into kaggle and give its path

M

Max•12/23/23, 10:27 PM

Oh, is that easy? Thank you, I appreciate it

A

AIGambino•12/23/23, 10:27 PM

A high-detail anime-style, especially attack on titan style, man cartoon character with brown hair, black eyes and big smile wearing comfortable, set in an environment with photorealistic and artistic details, enhanced by chiaroscuro lighting effects by bright sunshine. front view. Back ground with books.
Negative prompt: bad anatomy, bad hands, three hands, three legs, bad arms, missing legs, missing arms, poorly drawn face, bad face, fused face, cloned face, worst face, three crus, extra crus, fused crus, worst feet, three feet, fused feet, fused thigh, three thigh, fused thigh, extra thigh, worst thigh, missing fingers, extra fingers, ugly fingers, long fingers, horn, realistic photo, extra eyes, huge eyes, 2girl, amputation, disconnected limbs
Steps: 75, Sampler: DPM++ 2M SDE Karras, CFG scale: 6, Seed: 876854978, Size: 1673x696, Model hash: e8d456c42e, Model: toonyou_beta6, VAE hash: 63aeecb90f, VAE: vae-ft-mse-840000-ema-pruned.safetensors, Denoising strength: 1, ControlNet 0: "Module: none, Model: control_v11f1e_sd15_tile [a371b31b], Weight: 1, Resize Mode: Crop and Resize, Low Vram: False, Guidance Start: 0, Guidance End: 1, Pixel Perfect: True, Control Mode: ControlNet is more important, Save Detected Map: True", ControlNet 1: "Module: none, Model: diff_control_sd15_temporalnet_fp16 [adc6bd97], Weight: 1, Resize Mode: Crop and Resize, Low Vram: False, Guidance Start: 0, Guidance End: 1, Pixel Perfect: True, Control Mode: ControlNet is more important, Save Detected Map: True", ControlNet 2: "Module: softedge_pidinet, Model: control_v11p_sd15_softedge [a8575a2a], Weight: 1, Resize Mode: Crop and Resize, Low Vram: False, Processor Res: 512, Guidance Start: 0, Guidance End: 1, Pixel Perfect: True, Control Mode: ControlNet is more important, Save Detected Map: True", Pad conds: True, Version: v1.6.0-298-g5e80d9ee

A high-detail anime-style, especially attack on titan style, man cartoon character with brown hair, black eyes and big smile wearing comfortable, set in an environment with photorealistic and artistic details, enhanced by chiaroscuro lighting effects by bright sunshine. front view. Back ground with books.
Negative prompt: bad anatomy, bad hands, three hands, three legs, bad arms, missing legs, missing arms, poorly drawn face, bad face, fused face, cloned face, worst face, three crus, extra crus, fused crus, worst feet, three feet, fused feet, fused thigh, three thigh, fused thigh, extra thigh, worst thigh, missing fingers, extra fingers, ugly fingers, long fingers, horn, realistic photo, extra eyes, huge eyes, 2girl, amputation, disconnected limbs
Steps: 75, Sampler: DPM++ 2M SDE Karras, CFG scale: 6, Seed: 876854978, Size: 1673x696, Model hash: e8d456c42e, Model: toonyou_beta6, VAE hash: 63aeecb90f, VAE: vae-ft-mse-840000-ema-pruned.safetensors, Denoising strength: 1, ControlNet 0: "Module: none, Model: control_v11f1e_sd15_tile [a371b31b], Weight: 1, Resize Mode: Crop and Resize, Low Vram: False, Guidance Start: 0, Guidance End: 1, Pixel Perfect: True, Control Mode: ControlNet is more important, Save Detected Map: True", ControlNet 1: "Module: none, Model: diff_control_sd15_temporalnet_fp16 [adc6bd97], Weight: 1, Resize Mode: Crop and Resize, Low Vram: False, Guidance Start: 0, Guidance End: 1, Pixel Perfect: True, Control Mode: ControlNet is more important, Save Detected Map: True", ControlNet 2: "Module: softedge_pidinet, Model: control_v11p_sd15_softedge [a8575a2a], Weight: 1, Resize Mode: Crop and Resize, Low Vram: False, Processor Res: 512, Guidance Start: 0, Guidance End: 1, Pixel Perfect: True, Control Mode: ControlNet is more important, Save Detected Map: True", Pad conds: True, Version: v1.6.0-298-g5e80d9ee

A high-detail anime-style, especially attack on titan style, man cartoon character with brown hair, black eyes and big smile wearing comfortable, set in an environment with photorealistic and artistic details, enhanced by chiaroscuro lighting effects by bright sunshine. front view. Back ground with books.
Negative prompt: bad anatomy, bad hands, three hands, three legs, bad arms, missing legs, missing arms, poorly drawn face, bad face, fused face, cloned face, worst face, three crus, extra crus, fused crus, worst feet, three feet, fused feet, fused thigh, three thigh, fused thigh, extra thigh, worst thigh, missing fingers, extra fingers, ugly fingers, long fingers, horn, realistic photo, extra eyes, huge eyes, 2girl, amputation, disconnected limbs
Steps: 75, Sampler: DPM++ 2M SDE Karras, CFG scale: 6, Seed: 876854978, Size: 1673x696, Model hash: e8d456c42e, Model: toonyou_beta6, VAE hash: 63aeecb90f, VAE: vae-ft-mse-840000-ema-pruned.safetensors, Denoising strength: 1, ControlNet 0: "Module: none, Model: control_v11f1e_sd15_tile [a371b31b], Weight: 1, Resize Mode: Crop and Resize, Low Vram: False, Guidance Start: 0, Guidance End: 1, Pixel Perfect: True, Control Mode: ControlNet is more important, Save Detected Map: True", ControlNet 1: "Module: none, Model: diff_control_sd15_temporalnet_fp16 [adc6bd97], Weight: 1, Resize Mode: Crop and Resize, Low Vram: False, Guidance Start: 0, Guidance End: 1, Pixel Perfect: True, Control Mode: ControlNet is more important, Save Detected Map: True", ControlNet 2: "Module: softedge_pidinet, Model: control_v11p_sd15_softedge [a8575a2a], Weight: 1, Resize Mode: Crop and Resize, Low Vram: False, Processor Res: 512, Guidance Start: 0, Guidance End: 1, Pixel Perfect: True, Control Mode: ControlNet is more important, Save Detected Map: True", Pad conds: True, Version: v1.6.0-298-g5e80d9ee

A high-detail anime-style, especially attack on titan style, man cartoon character with brown hair, black eyes and big smile wearing comfortable, set in an environment with photorealistic and artistic details, enhanced by chiaroscuro lighting effects by bright sunshine. front view. Back ground with books.
Negative prompt: bad anatomy, bad hands, three hands, three legs, bad arms, missing legs, missing arms, poorly drawn face, bad face, fused face, cloned face, worst face, three crus, extra crus, fused crus, worst feet, three feet, fused feet, fused thigh, three thigh, fused thigh, extra thigh, worst thigh, missing fingers, extra fingers, ugly fingers, long fingers, horn, realistic photo, extra eyes, huge eyes, 2girl, amputation, disconnected limbs
Steps: 75, Sampler: DPM++ 2M SDE Karras, CFG scale: 6, Seed: 876854978, Size: 1673x696, Model hash: e8d456c42e, Model: toonyou_beta6, VAE hash: 63aeecb90f, VAE: vae-ft-mse-840000-ema-pruned.safetensors, Denoising strength: 1, ControlNet 0: "Module: none, Model: control_v11f1e_sd15_tile [a371b31b], Weight: 1, Resize Mode: Crop and Resize, Low Vram: False, Guidance Start: 0, Guidance End: 1, Pixel Perfect: True, Control Mode: ControlNet is more important, Save Detected Map: True", ControlNet 1: "Module: none, Model: diff_control_sd15_temporalnet_fp16 [adc6bd97], Weight: 1, Resize Mode: Crop and Resize, Low Vram: False, Guidance Start: 0, Guidance End: 1, Pixel Perfect: True, Control Mode: ControlNet is more important, Save Detected Map: True", ControlNet 2: "Module: softedge_pidinet, Model: control_v11p_sd15_softedge [a8575a2a], Weight: 1, Resize Mode: Crop and Resize, Low Vram: False, Processor Res: 512, Guidance Start: 0, Guidance End: 1, Pixel Perfect: True, Control Mode: ControlNet is more important, Save Detected Map: True", Pad conds: True, Version: v1.6.0-298-g5e80d9ee

settings

A

AIGambino•12/23/23, 10:27 PM

PPflanzen die Tanzen https://www.youtube.com/watch?v=vi6MDIt9P0c

F

Furkan Gözükara SECourses•12/23/23, 10:27 PM

nice. perhaps embedding those fancy subtitles may help

A

AIGambino•12/23/23, 10:27 PM

Any idea why img2img just doesn't change much? I've tried changing seeds, cfg and even the model, but never get a comic book or anime style face no matter the prompt or anything

AAIGambino Any idea why img2img just doesn't change much? I've tried changing seeds, cfg an...

F

Furkan Gözükara SECourses•12/23/23, 10:28 PM

probably controlnet instead of img2img

F

Furkan Gözükara SECourses•12/23/23, 10:28 PM

would work better

A

AIGambino•12/23/23, 10:28 PM

I have controlnet enabled, but I'm running batches, would that still work?

AAIGambino I have controlnet enabled, but I'm running batches, would that still work?

F

Furkan Gözükara SECourses•12/23/23, 10:29 PM

i think should

F

Furkan Gözükara SECourses•12/23/23, 10:29 PM

not sure search for it

FFurkan Gözükara SECourses nice. perhaps embedding those fancy subtitles may help

P

Pflanzen die Tanzen•12/23/23, 10:34 PM

this was just a simple test on some training data preps. Testing an idea, which "sort of" worked.
But thx for the feedback, will consider in future tests.
Currently the focus is dataset preparations.

AAIGambino ```A high-detail anime-style, especially attack on titan style, man cartoon char...

X

xlads•12/24/23, 1:12 AM

Looks awesome

F

Furkan Gözükara SECourses•12/24/23, 1:45 AM

training myself on an anime SDXL model

F

Furkan Gözükara SECourses•12/24/23, 1:45 AM

lets see outputs

F

Furkan Gözükara SECourses•12/24/23, 1:45 AM

by the way check this out if you havent yet : https://www.reddit.com/r/StableDiffusion/comments/18pdfnc/tested_275_sdxl_prompt_styles_of_fooocus_on/

From the StableDiffusion community on Reddit: Tested 275 SDXL Promp...

Explore this post and more from the StableDiffusion community

M

Max•12/24/23, 10:12 AM

Hey guys, does anyone know how to install tensorRT and adetailer on kaggle?

T

theonlyjoshyouknow•12/24/23, 11:06 AM

@Dr. Furkan Gözükara I've noticed that for your most recent tests you used Koya SS GUI for training, but on Patreon I've only been able to find setup scripts for the Dreambooth extension on RunPod. I don't see any reason I couldn't manually run Koya on a RunPod instance, but I'm hesitant because of the issues with Dreambooth only working with specific versions. Does Koya have similar issues? Also, what advantages does Koya have for training on individual people compared to the A1111 Dreambooth extension?

T

theonlyjoshyouknow•12/24/23, 11:06 AM

So far every model I've trained (using RealisticVision as a base) has given me outputs that are consistently blurry