- Dec 31, 2016
- 316
- 708
Some other use for [] can be found
You must be registered to see the links
- Dec 23, 2018
- 506
- 2,095
Can I recommend Styles? I use the ones from this free Patreon post :-
Makes some interesting outputs from the same basic prompt. Here's some examples from the same model and prompt (different seeds though)
Vector Illustrations
then Indie Game
and finally Black and White
You can get some awesome outputs just by adding a style, and of course can make your own
You must be registered to see the links
Makes some interesting outputs from the same basic prompt. Here's some examples from the same model and prompt (different seeds though)
Vector Illustrations
then Indie Game
and finally Black and White
You can get some awesome outputs just by adding a style, and of course can make your own
- Dec 23, 2018
- 506
- 2,095
So I thought I'd do a quick exercise to show how models/checkpoints/styles can change the entire look of the finished pic. I took the below prompt :-
1girl, nsfw, full body, (masterpiece, best quality, ultra-detailed, best shadow), full body, freedom, soul, cyberpunk, perfect anatomy, centered, approaching perfection, dynamic, highly detailed, smooth, sharp focus,
Negative prompt: canvas frame, (high contrast:1.2), (over saturated:1.2), (glossy:1.1), cartoon, 3d, ((disfigured)), ((bad art)), ((b&w)), blurry, ((bad anatomy)), (((bad proportions))), ((extra limbs)), cloned face, (((disfigured))), extra limbs, (bad anatomy), gross proportions, (malformed limbs), ((missing arms)), ((missing legs)), (((extra arms))), (((extra legs))), mutated hands, (fused fingers), (too many fingers), (((long neck))), Photoshop, video game, ugly, tiling, poorly drawn hands, 3d render
Steps: 50, Sampler: DPM++ 2M Karras, CFG scale: 7, Seed: 3757479247, Size: 544x960, Model hash: 03363589fe, Model: cyberrealistic_v31, Version: v1.3.2
and thought I would run it through a plot to show how it looks with different models etc. Warning! Large pic
Some interesting outcomes. RPG v4 is set up to do d&d characters in armour etc, which is why I think some of its outcomes are a bit weird.
1girl, nsfw, full body, (masterpiece, best quality, ultra-detailed, best shadow), full body, freedom, soul, cyberpunk, perfect anatomy, centered, approaching perfection, dynamic, highly detailed, smooth, sharp focus,
Negative prompt: canvas frame, (high contrast:1.2), (over saturated:1.2), (glossy:1.1), cartoon, 3d, ((disfigured)), ((bad art)), ((b&w)), blurry, ((bad anatomy)), (((bad proportions))), ((extra limbs)), cloned face, (((disfigured))), extra limbs, (bad anatomy), gross proportions, (malformed limbs), ((missing arms)), ((missing legs)), (((extra arms))), (((extra legs))), mutated hands, (fused fingers), (too many fingers), (((long neck))), Photoshop, video game, ugly, tiling, poorly drawn hands, 3d render
Steps: 50, Sampler: DPM++ 2M Karras, CFG scale: 7, Seed: 3757479247, Size: 544x960, Model hash: 03363589fe, Model: cyberrealistic_v31, Version: v1.3.2
and thought I would run it through a plot to show how it looks with different models etc. Warning! Large pic
You don't have permission to view the spoiler content.
Log in or register now.
Some interesting outcomes. RPG v4 is set up to do d&d characters in armour etc, which is why I think some of its outcomes are a bit weird.
- Sep 15, 2021
- 888
- 3,274
"Quick" exercise? That must have taken a very long time to generate!
Thank you for doing this, X/Y/Z plots are so useful (if time consuming) to visualise the effects of changing parameters.
Thank you for doing this, X/Y/Z plots are so useful (if time consuming) to visualise the effects of changing parameters.
- Mar 17, 2021
- 301
- 1,029
nice, i like see these test. it is really useful. thank you so much.
- Dec 21, 2020
- 1,526
- 3,596
The openpose controlnet is savage. I couldn't make myself set it up in ComfyUI because lazy, but once I did, I can't get enough.
You don't have permission to view the spoiler content.
Log in or register now.
- Jan 10, 2019
- 268
- 1,435
Has anyone ever had such colour 'bleeds' or 'patches' in their generated images? I've had the issue today, restarted SD, updated it, but still there in all models tested. Use VAE/no VAE, hires or not, tried different settings. No idea where this comes from. Any help would be very welcome.
- Sep 15, 2021
- 888
- 3,274
I have seen the "colour patches" before, from an over-baked TI being applied at too high a strength. Maybe try reducing the strength a little (Patricia37:0.6 perhaps?)
Also, you've got this in the prompts:
"rich colours hyper realistic lifelike texture dramatic lighting unreal engine trending on artstation cinestill 800"
Without commas or full stops it could be a little unclear how SD's supposed to apply these.
Of course, I could be wrong, it could be something else, but your CFG isn't excessive, everything else looks OK too.
- Dec 21, 2020
- 1,526
- 3,596
My bet is there's not enough steps for the renderer to completely flashout the image. Do humor me - please put 20 steps and see if this goes away.
- Sep 15, 2021
- 888
- 3,274
Apologies, I missed that - 15 steps is probably not enough, even with 30 Hi-res steps.
BTW, I'm currently running a few tests to see what the effects of more steps and perhaps other things are. I don't have the "Patricia37" TI as it's a personally-created one of sharlotte's , so that will inherently remove any effects from it from the running.
Last edited:
- Sep 15, 2021
- 888
- 3,274
OK, here's a test with 10, 15, 20, 25 & 30 steps*:
10:
15:
20:
(notably, the patches were still visible until the upscaling got rid of them)
25:
30:
Grid:
My personal preference here would be 30 steps, but 25 works well too.
As I have previously said: "Play around with more or fewer steps to see what works best for what you're aiming for, but be prepared to increase or decrease them at any time"
*Without the TI as it's not publicly available.
10:
15:
20:
(notably, the patches were still visible until the upscaling got rid of them)25:
30:
Grid:
My personal preference here would be 30 steps, but 25 works well too.
As I have previously said: "Play around with more or fewer steps to see what works best for what you're aiming for, but be prepared to increase or decrease them at any time"
*Without the TI as it's not publicly available.
- Sep 26, 2018
- 3,040
- 3,134
I'm having that problem as well and I also having the faces deformed and it doesn't matter if I have deformed_face in the negative prompt
- Jan 24, 2020
- 1,401
- 3,793
I agree with both Jim and Seph. Increase the sample steps and make sure to use Face restoration and/or post processing.
The next thing is to take a look at the checkpoint and/or Lora's and TI's etc that could be the cause.
- Dec 21, 2020
- 1,526
- 3,596
It would be easier for us to help you if you posted a file with the render.
- Sep 26, 2018
- 3,040
- 3,134
I always have at 48 and above
i am in processes of redownloading it
here is a face problem this one is at 20
this one is at 48
I can't get the other problem to occur after re-downloading it
Last edited:
- Jan 10, 2019
- 268
- 1,435
Thanks a mil guys, I'll try later this evening and will up the number of steps, insert proper commas and lower the strength of the TI and see where that takes me. All these make perfect sense. Thanks for the quick turnaround!
- Sep 15, 2021
- 888
- 3,274
I really would like to see the results (and I know for sure Sepheyer does, these girls are very much his type!)
- Sep 15, 2021
- 888
- 3,274
OK, couple of recommendations:
- In Settings, make sure you're using GFPGAN for Restore Faces, not the default CodeFormers.
- Enable Hi-res Fix. If your GPU doesn't really have the horsepower for it, just set it as low as you can so it adds at least 8 pixels in each direction. As a minimum use the same number of hires steps as gen steps, ideally double. Don't use Latent, it just blurs stuff. Experiment with which works best for you.
- Mar 17, 2021
- 301
- 1,029
read some answer here, all good, but what sampler? some sampler need more step, or you get these patch.
Also some lora generate this, for example i have some lora and have to weight it to 0.2 for get good result( a bit overtrained i think).
- Jan 24, 2020
- 1,401
- 3,793
Generating images with multiple subjects is notoriously problematic without help from a Lora or similar. I would try out a Lora that has this specific concept (multiple girls) and try with the bronze statue prompt.
Like this one.
Example:
Source:
You must be registered to see the links