me3
Member
- Dec 31, 2016
- 316
- 708
Not recently, but it works fairly well.Have you tried latent composition? It sounds interesting.I have seen it here and there at a glance but not tried it.
However you need to be using prompts that "fit" together and just do a very small separate sampling before putting things together. So you won't really know how element look until you're done.
For comfyui users this can serve as
You must be registered to see the links
, not sure how you'd do this in a1111, workflow is in the images. For those that don't use comfy you can upload the image to
You must be registered to see the links
, scroll down about half the page and you should see a box mentioning "drag and drop ... png or json"
Jimwalrus
Well-Known Member
- Sep 15, 2021
- 1,118
- 4,260
If you've created TIs for each face, then you can use the Embedding Inspector to combine them at whatever relative strengths you wish (+ a dash of, for instance, Ariana Grande if you like).
If you've created LoRAs for each of them, I believe Kohya_ss has a LoRA combination tool.
Thanks. I'm not that advanced yet. I used img2img, and slowly blended them together for a decent result
me3
Member
- Dec 31, 2016
- 316
- 708
Just to followup my previous post
This just a VERY quick mockup test following the layout from the link in that post and sticking with the cabaret interest that started it
I should have given a bit more thought to the model i picked and the prompts were just "stage at moulin rouge" and "female cabaret dancer at moulin rouge" with the usual quality pos and negatives. As it's just simple SD15 i did run it through a face detailing when upscaling.
There's some problems but they are by far more affected by lacks of prompting and tweaking settings, than the method itself.
This just a VERY quick mockup test following the layout from the link in that post and sticking with the cabaret interest that started it
I should have given a bit more thought to the model i picked and the prompts were just "stage at moulin rouge" and "female cabaret dancer at moulin rouge" with the usual quality pos and negatives. As it's just simple SD15 i did run it through a face detailing when upscaling.
There's some problems but they are by far more affected by lacks of prompting and tweaking settings, than the method itself.
devilkkw
Member
- Mar 17, 2021
- 338
- 1,146
Lora is better way, more controllable than embedding. what problem have you on training lora? if you expose maybe we give better help.
Also, for lora, you have 2 way to chose: train directly mixed, or train 2 model and merge them.
devilkkw
Member
- Mar 17, 2021
- 338
- 1,146
Testing all 22 sampler in CUI, same settings, changing only sampler.
As you can see, LCM sampler is really different, all other similar.
Hope this comparison is useful, sorry for not attaching workflow on image, but is all messed up with node i'm experimenting and not good for sharing at the moment.
As you can see, LCM sampler is really different, all other similar.
You don't have permission to view the spoiler content.
Log in or register now.
You don't have permission to view the spoiler content.
Log in or register now.
You don't have permission to view the spoiler content.
Log in or register now.
You don't have permission to view the spoiler content.
Log in or register now.
Hey thanks for the support. I don't think my video card has enough juice to train a lora.. or i'm not doing it properly..
I've watched a few walkthrough videos, but I can't seem to figure it out..
Mr-Fox
Well-Known Member
- Jan 24, 2020
- 1,401
- 3,808
That's some mighty fine Wendigo Erotica.
A little tip or just observation, the LCM sampler will not give a good result if you are not using it with a LCM checkpoint. Also, if you use a resolution 1024 wih SD1.5 you are more likely to get conjoined twins. I would recommend to use 960x640 and then use either hiresfix or upscale in img2img with SD Upscale script. I know for a fact that you are already aware. This is only a reminder and for anyone else that might not be aware.
Last edited:
Mr-Fox
Well-Known Member
- Jan 24, 2020
- 1,401
- 3,808
Jimwalrus
Well-Known Member
- Sep 15, 2021
- 1,118
- 4,260
Civitai has an
You must be registered to see the links
. It's not perfect as you don't have total control, but I've used it a few times. Costs 500 'Buzz' each time (even failures), so either get liking some other peoples' images or let me know your ID there and I'll tip you 1000 to let you have a play with it.
Mr-Fox
Well-Known Member
- Jan 24, 2020
- 1,401
- 3,808
There are not that many LCM checkpoints available as of yet compared to "normal" SD1.5 and SDXL, though indeed a few.
The point of LCM (Latent Consistency Model) is to be able to run fewer steps and lower cfg scale to cut down on generation time but still get a high quality.
The rule of thumb is 6-12 steps and 1-4 cfg scale. 10 steps and 1-2 cfg scale seems to be good with most models.
I ran a few ckpt compare tests with plot script and I borrowed the prompt of the great Devilkkw's delicious Cryptid Babe.
SD1.5 LCM 1024x1280 (notice the tendency for conjoined twins):
SD1.5 LCM 640x960 (notice the absence of conjoined twins):
There are also XL LCM models. As most know you can use a higher resolution with XL models.
The rule of thumb is that the resolution should be equal to 2048. You can try different ratios. One that I have found to work well for me is 896x1152.
(Thanks to the eminent Synalon for providing this list).
SDXL LCM 896X1152:
A tip is to never use the standard VAE that was released with the first SDXL model, it's slow as hell.
I recommend fenrisxl VAE instead, it's faster. SDXL LCM is still much slower in general though compared to normal SD1.5 or LCM, at least with an older GPU like the 1070 card I have.
SDXL LCM with fenrisxl VAE 896X1152:
You must be registered to see the links
The point of LCM (Latent Consistency Model) is to be able to run fewer steps and lower cfg scale to cut down on generation time but still get a high quality.
The rule of thumb is 6-12 steps and 1-4 cfg scale. 10 steps and 1-2 cfg scale seems to be good with most models.
I ran a few ckpt compare tests with plot script and I borrowed the prompt of the great Devilkkw's delicious Cryptid Babe.
SD1.5 LCM 1024x1280 (notice the tendency for conjoined twins):
SD1.5 LCM 640x960 (notice the absence of conjoined twins):
There are also XL LCM models. As most know you can use a higher resolution with XL models.
The rule of thumb is that the resolution should be equal to 2048. You can try different ratios. One that I have found to work well for me is 896x1152.
(Thanks to the eminent Synalon for providing this list).
SDXL LCM 896X1152:
A tip is to never use the standard VAE that was released with the first SDXL model, it's slow as hell.
I recommend fenrisxl VAE instead, it's faster. SDXL LCM is still much slower in general though compared to normal SD1.5 or LCM, at least with an older GPU like the 1070 card I have.
You must be registered to see the links
SDXL LCM with fenrisxl VAE 896X1152:
Last edited:
Mr-Fox
Well-Known Member
- Jan 24, 2020
- 1,401
- 3,808
Bonus.
(twigs and pine cones included)..
I hope the excellent Devilkkw doesn't mind I keep posting with his prompt, had too much fun to stop.
The other day me and the eminent Synalon experimented with and explored variation seed, variation strength and more importantly variation resolution.
This is what the tip text says about it:
This means that you can generate an image in landscape without getting eldritch monsters by setting the main resolution in landscape ratio and variation resolution in portrait ratio. See examples below.
(twigs and pine cones included)..
I hope the excellent Devilkkw doesn't mind I keep posting with his prompt, had too much fun to stop.
The other day me and the eminent Synalon experimented with and explored variation seed, variation strength and more importantly variation resolution.
This is what the tip text says about it:
This means that you can generate an image in landscape without getting eldritch monsters by setting the main resolution in landscape ratio and variation resolution in portrait ratio. See examples below.
Last edited:
me3
Member
- Dec 31, 2016
- 316
- 708
How you describe the variation resolution sounds similar to how you can use width/height and target width/target height in sdxl text encoder in comfyui. If that's the case it's very useful, since among many things you can "zoom" in/out or generation, and decide what gets "cropped out" and how it fits on your "canvas"Bonus.
(twigs and pine cones included)..
I hope the excellent Devilkkw doesn't mind I keep posting with his prompt, had too much fun to stop.
The other day me and the eminent Synalon experimented with and explored variation seed, variation strength and more importantly variation resolution.
This is what the tip text says about it:
View attachment 3268917
This means that you can generate an image in landscape without getting eldritch monsters by setting the main resolution in landscape ratio and the variation resolution in portrait ratio. See examples below.
View attachment 3268862 View attachment 3268863
View attachment 3268882 View attachment 3268928
Regarding LCM, you don't have to have a model trained for it, there are LCM weight loras, for both SD15 and XL, which lets you use any model and you can create images at less steps and cfg. You use them as any other lora and they work pretty well.
You must be registered to see the links
You must be registered to see the links
There's a Lora for SDXL Turbo too
You must be registered to see the links
And one that combine both LCM and Turbo
You must be registered to see the links
Edit:
Since i had to go look for the vae Mr-Fox mention (compulsive need to try new things), it wasn't that easy to find right away, too many model version listed, so here's
You must be registered to see the links
Last edited:
Mr-Fox
Well-Known Member
- Jan 24, 2020
- 1,401
- 3,808
My bad. I have updated the post with the link. Should not be an issue now that we both have linked to it.
Btw, the FenrisXL model itself is also excellent. Now this can't possibly need it's own link..
But here it is anyway:
You must be registered to see the links
Image Example:
https://f95zone.to/threads/ai-art-show-us-your-ai-skill-no-teens.138575/post-11967954
Last edited:
devilkkw
Member
- Mar 17, 2021
- 338
- 1,146
Oh, never see lcm models. And never try it, is possible to port standard .safetensors to lcm? benefit?That's some mighty fine Wendigo Erotica.
A good reminder you say Mr-Fox, sd 1.5 model work great with low res, and push it high is really a pain in the ass, many models get double and weird result at 768, so generate and upscale seem good solution.
But i have to ask a question: i use my model merged many times with merge block weight in a111, to push out resolution, but in a1111 max resolution reach 896x1152 and in CUI i reach 1024x1280. why so much different?
I also checked sampling method code, and seem work different in CUI and in a1111, but if the sampler is the same, why?
Wow, love these type of post, is really useful for me to have a good idea on how it work.thank you.
A ot question: how many checkpoint you have?
So beautiful result's. i'm glade you used my prompt for sample. And good test, variation seed is bit underestimate,keep testing and share. I'm really interested on it.Bonus.
(twigs and pine cones included)..
I hope the excellent Devilkkw doesn't mind I keep posting with his prompt, had too much fun to stop.
The other day me and the eminent Synalon experimented with and explored variation seed, variation strength and more importantly variation resolution.
This is what the tip text says about it:
View attachment 3268917
This means that you can generate an image in landscape without getting eldritch monsters by setting the main resolution in landscape ratio and variation resolution in portrait ratio. See examples below.
View attachment 3268862 View attachment 3268863
View attachment 3268882 View attachment 3268928
Too general, what are your pc spec? what are you using for train? video driver version?
Also read some post ahead, other user give you a suggestion on a possible alternative way to train.
lobotomist
Well-Known Member
- Sep 4, 2017
- 1,009
- 1,081
sorry for the noob question, i have an intel card and i know that you can use auto 1111 with open vino. but what about comfy UI?
Also other recommendations for beggkners with intel cards?
Also other recommendations for beggkners with intel cards?
Mr-Fox
Well-Known Member
- Jan 24, 2020
- 1,401
- 3,808
I have no idea.
The benifit is like I said in my post to be able to use much less steps and lower cfg scale to cut down on generation time and still get high quality.
As I don't use the spagetti ui I can't help you with ComfyUi.
With SD1.5 it's best to keep it under 1024 in either direction. So I use 640x960 for portrait ratio, I simply flip it when I do landscape. I use this resolution while searching for a good seed and then when I found it
I then re-use that seed and enable hiresfix with 2x upscale and a fairly low denoising to get a sharp image. Then I might upscale it further with SD upscale script in img2img.
Way too many probably..Wow, love these type of post, is really useful for me to have a good idea on how it work.thank you.
A ot question: how many checkpoint you have?
I'm glad you liked it.
Last edited:
I'm going to test training sdxl on pornographic concept by using color association to ease the formation of the neural network. I don't know anything about that but I assume it creates associations so it should work.
Essentially, I'll separate the image into two identical images, then color specific regions. Then I'll prompt what the colors are associated to. The AI knows what the colors are, so it will associate it with the concept to learn.
Here are some example images.
Essentially, I'll separate the image into two identical images, then color specific regions. Then I'll prompt what the colors are associated to. The AI knows what the colors are, so it will associate it with the concept to learn.
Here are some example images.
Last edited: