Another new day a new resumption

[cerberus]

I succeed to start the program but I have a weak laptop and a strong laptop, I use gencraft on weak laptop, my system freezes when trying to create a image so I will try to use my strong laptop.

Thanks for your info Cerberus, I really appreciate that!

I'm not sure how fast Stable Diffusion will be on a laptop, it's optimised for use with a dedicated graphics card (it takes advantage the parallel processing specialization and faster memory) but I think you can run it. I seem to remember the video mentions options for running with low memory.

Sometimes SD can look like it is frozen, but check the CMD window before killing it. If you are doing something for the first time it may be downloading additional files from the internet (e.g. samplers, schedulers, upscalers etc), you can see if it is doing this in the CMD window, there will be a progress bar and numbers indicating how much has been downloaded.

Initially I'd stick with the SD 1.5 checkpoint models, (these are about 2GB each) and avoid the SDXL models (6.5GB). SD 1.5 used less resources and aligns with the prompts better. My original thread is all SD 1.5 images and starts with me in the position you're in now:

viewtopic.php?f=30&t=27259

I don't know what checkpoint model you've chosen, I'd recommend AbsoluteReality, DreamShaper and/or EpicRealism as good ones to start with. I'd also suggest you add an extension (you do this through the "Extensions" tab) called ADetailer, which has options that can fix some of the face issue you get starting out. I can give you some pointers when you are ready to try this out.

[Scarol]

I managed to use the program on my stronger laptop, but the program is too extensive for me, so I prefer to myself to use Gencraft, unfortunately Gencraft does not always do what I describe and what I want to see, Gencraft is sometimes stubborn.

Thanks again for your help Cerberus and I’m looking very forward to your new photos!

[cerberus]

I managed to use the program on my stronger laptop, but the program is too extensive for me, so I prefer to myself to use Gencraft, unfortunately Gencraft does not always do what I describe and what I want to see, Gencraft is sometimes stubborn.

Thanks again for your help Cerberus and I’m looking very forward to your new photos!

I wouldn't get rid of SD just yet, you don't have to learn it all in a day. Keep it around and play with it from time to time. There are a few things that are worth knowing that you may not have found yet.

- All images generated go in the "outputs" folder in the stable diffusion folder.
- If you right click on the "Generate" button you will see an option "Generate Forever", this causes SD to run until you click the "Cancel Generate Forever" or you kill the session. So you can leave it to run unattended and check the output later.
- If you click the "Dice" by the seed number it sets it to -1 (i.e. random selection), using this and "generate forever" you can leave it to generate a series of different images.
- There are extensions for wildcards and dynamic prompts, which change the prompt each cycle. This is how I use SD.

One other thing that you may find useful is a little stand-alone app called "SD Prompt Reader", this is a freestanding program that displays the metadata associated with SD generated images. If you install this you can save my SD images locally and open them with SD Prompt Reader, which will show you the settings I used to generate the image. This includes the prompt, negative prompt and all the other settings. So it you see one of my images you like you can try the same prompt/Nprompt in Gencraft and see what that produces. Note: if I've edited my image in photoshop (which I do sometimes to fix hands etc) you won't find any metadata, photoshop strips it out, but there aren't many images like this.

[cerberus]

or for motivation, i could leave and not come back?
(i'd come back eventually, but thinking i'm gone would motivate you, right?)

easy knots are no challenge, you'll appreciate this... later

Naw, she's a pro she'd get out on her own and come looking for you.

When you are up to your tits in a bottomless bog there are no such things as "easy knots"!

Tonight we have:

"Hmm, I guess this is why he laughed when he pointed to this spot on the map and said - yes that's where it is!"

00517-2951178545.png

"It's somewhere around here, I'm sure of that!"

00529-446065294.png

"Shit, he was here a moment ago, where did he go?"

00539-97158706.png

Slow and steady does it, no need to rush!

00552-2002665004.png

"Woah, sudden drop there!"

00254-3734224101.png

[cerberus]

Some more:

"For some reason it seems a bit claustrophobic in here!"

00626-3343741817.png

"Yep, he's gone, the two-timing bastard!"

00627-2526825005.png

Uh-oh, my boots are flooded, that wasn't supposed to happen? Right RIGHT!"

00637-1314864086.png

"Um swimming movements don't seem to be working. What else you got?"

00673-1309608975.png

"Perhaps I shouldn't have jumped off that log!"

00735-802508111.png

[Scarol]

I like the soldier one very well!

[MadMax359]

right coordinates... maybe 3 feet down?

you had me at Slow and Steady

the best way to fight claustrophobia is to take deep breaths...

[Scarol]

If I may post this picture here..
I have this screen, I have no idea what to do now...

For Ceberus.jpg

[cerberus]

If I may post this picture here..
I have this screen, I have no idea what to do now...

For Ceberus.jpg

You are in the Img2img tab, this is used to generate an image based on another image. If you click in the white space in the bottom left of your screen print you'll get a dialogue to load an image from your PC. If you then click "Generate" you'll probably get a fairly unpleasant version of the source image displayed below the Generate button. On your screen below the source image are various settings used to create the new image. You'll need to set some of these to start getting reasonable results. I don't know the model you have selected (top left) so I can only make suggestions but experiment with the "sampler" (Try "Euler a", it's reasonably flexible), "scheduler" (Karras or SGM Uniform maybe), number of steps (20 is a good starting point}. Start with a CFG Scale of 1.5 and Denoise Strength of .15 and increase to see if things improve. You can put something in the prompt to help with quality eg "high quality photography, 8k, hdr, sharp focus, depth of field, professional photograph, super-detailed, high resolution, award winning photo, bokeh, smooth". This stuff is usually put it a style file but you probably don't have one of those yet.

The error message you were getting was because you hadn't set up a source image etc.

Try this and see how that goes.

If I were you I'd start with the Txt2img tab and generate images from a prompt. Have you downloaded any other models from Civitai yet? The model your screen print shows isn't one of the top ones.

[bogbud]

I'm back and right in time to keep on commenting before there are too many of those godies!

She's such a tease!
00036-1307123728.png

Nice one. Also hands are drawn really good!

"Um swimming movements don't seem to be working. What else you got?"
00673-1309608975.png

That ditch in this bog is quite nasty. Just put your arms right next to your body and then try to lift as much mud up as possible. Also try to bend and pump your legs simultaneously . Lets see if that helps, in any case i can search for a stick later....