Draw things is great. Really, really great. I'm kind of amazed that there's so little information about it around. It took some time for me to get a handle of exactly what was going on and what each module and setting was actually doing. A bit like the Blender learning curve; you're lost at sea until you get a grip on the workflow, then you're off to the races. Wish I had seen this first. Nice work. Thanks.
I'll see how this video does for a bit, but if I do this, I'd have it on Patreon and possibly Udemy, TH-cam doesn't know what to do with these tutorials.
IMPORTANT TO NOTE: If you have concerns about your physical storage. This WILL take up residency on macOS Sonoma. It stores this information locally, thou hidden in the following folder(s): Users-> [THE ACCOUNT NAME]->Library->Containers->Draw Things
Thanks, my previous AI art video was a slow burn, it has like 20k views and as a smaller channel that’s ok. The other video shows multiple apps but this is the only one anyone should bother with. I think TH-cam doesn’t know what to do with this content as I’m not an AI channel even though I have three AI vids.
New sub here-thank you for including iPad ai art generation as I bought it for art but the programmers seem very few considering the hype for iPad being the next big thing and it is the best but seems to be ignored for ai art generators. I thought OpenAI was wow until MS/BG took over control with censorship which they’re doing everywhere like what happened to perplexityai when co pilot got added. So I’ve been searching for a way to stay free artistically. Your tip for using pixelmator was something I had never heard about before and a great find for fixing ai art. More iPad tutorials please???
What would you like to see? I’ve been thinking about iPad content, I think as far as AI art goes I covered enough for the iPad itself as the more advanced stuff really is best done on mac.
FYI, it’s pretty easy to end up in some racy material. The AI crowd likes a lot of questionable material. If you go this route, I’d limit it to the models from Stable Diffusion as they’re responsible, on many fronts. It’s not always clear what models can do, in the SDXL model students could try to generate NSFW but would be severely disappointed.
OMG THANK YOU SO MUCH FOR SHOWING SOME TUTORIALS TOOOO!!! dude the only tutorial about this app i could find on youtube was in Chinese and i don't understand a bit 😭
Try TH-cam thumbnail title like ai art generator to download for iPad specifically because that’s what it’s really for with the pencil there is no other TH-cam video for iPad with SDXL download to iPad for creators-or in my case wanna be creator.
Even tho im a Windows user, i love your vids, are sooo satisfying to watch and also very helpful (Probably im getting an Intel iMac to use like a secondary PC)
@@tamalchakraborty5346 is an iMac, not the Mac Pro But just in case, the one im probably getting, has a Radeon 580 Pro, imma do a dual boot to Windows too
I thought it odd that you used the SDXL refiner model as your base model, instead of the two part base/refiner workflow it was intended for. I hadn’t seen that before!
Its the one that DrawThings vectors you to, it's either a merge or downloads the base in tandem (I can't recall off the top of my head). Trying to explain checkpoints and such in this video would have been information overload.
Not a ton to say, just expansion packs to achieve a desired outcome: art style, pose, character, object, setting etc. If you tried to use one designed for photo realism on a model designed to mimic impressionist painting it’s might not work well. It’s a bit of a grab bag. I’d read on AI websites if you want a technical or deep dive, or quiz chat gpt about it.
For some reason I'm getting the first image behind my generated image when I'm following along, and I don't see any tool to get rid of this. It's weird. The zoom with the wheel on the mouse is almost impossible to use because it's so sensitive. I guess there's no way to zoom in without using that.
Hi. I have been having a play around with Draw things & installed some other AI models. I was just wondering if you know of the "Photoshop" of AI text to image generator tools? As I notice that draw things is unhappy, importing some the bigger (over 4GB) AI models out there and I wondered if their was another tool designed for bigger AI models.
The limitations of larger models is a function of hardware right now. If you're on a lower end or older Mac, then that'll affect it. I've had zero issues with large 12 GB models but I'm also running it primarily on a Mac Pro 2019 with Radeon 6900 XT and M1 Max with 32 GB of RAM. I'm pretty positive there won't be anything with remotely the polish of Photoshop besides using Photoshop itself with it's own AI tools, at least not for many years. GIMP still feels like a beta of Photoshop 4.0 and that's the best-of for OpenSource image editing. Plus, the technology is young, the method of image generation imprecise, the changes still piling in (see SD1.0 vs SD1.5 vs SD2.0 vs SDXL) and the tools esoteric like poses. The best you'll see is pane-based interfaces like we have right now as that's the case of paid apps that have image generation. They're often only slightly more polished interface wise and much-more featured limited.
The best you’ll find is upscaling, using things like Topaz Photo or Pixelmator Pro’s AI upscaler or some of the web utilities. Unfortunately there’s a real limit to these tools as they can’t infer data that isn’t there so unless you have decent quality reference sources, these tools aren’t super useful. I don’t cover these techniques in this video but I’d try a demo of topaz.
Bro, you go that fast. I can barely tell what you’re clicking. You said. Press the camera icon. You did it at this rate which only the gods could see.😂
There are stable diffusion video models but the results are pretty bad, you can generate either 14 or 28 frames with the two models. It takes forever, and even using tricks like upscaling, and frame generation in other apps to improve the frame rate, videos are mostly psychedelic smeary messes that are 1-5s long. There’s a few online demos you can play with that are free to use, which do as good of a job (and are faster).
Draw things is great. Really, really great. I'm kind of amazed that there's so little information about it around. It took some time for me to get a handle of exactly what was going on and what each module and setting was actually doing. A bit like the Blender learning curve; you're lost at sea until you get a grip on the workflow, then you're off to the races. Wish I had seen this first. Nice work. Thanks.
Yes I would be interested in follow up videos regarding draw things
I'll see how this video does for a bit, but if I do this, I'd have it on Patreon and possibly Udemy, TH-cam doesn't know what to do with these tutorials.
IMPORTANT TO NOTE: If you have concerns about your physical storage. This WILL take up residency on macOS Sonoma. It stores this information locally, thou hidden in the following folder(s): Users-> [THE ACCOUNT NAME]->Library->Containers->Draw Things
Yeah, good point on managing the models. Probably should have tossed that in.
No worries, I enjoy your content!
This video needs to blow up! Subscribed!
Thanks, my previous AI art video was a slow burn, it has like 20k views and as a smaller channel that’s ok.
The other video shows multiple apps but this is the only one anyone should bother with. I think TH-cam doesn’t know what to do with this content as I’m not an AI channel even though I have three AI vids.
New sub here-thank you for including iPad ai art generation as I bought it for art but the programmers seem very few considering the hype for iPad being the next big thing and it is the best but seems to be ignored for ai art generators. I thought OpenAI was wow until MS/BG took over control with censorship which they’re doing everywhere like what happened to perplexityai when co pilot got added. So I’ve been searching for a way to stay free artistically. Your tip for using pixelmator was something I had never heard about before and a great find for fixing ai art. More iPad tutorials please???
What would you like to see? I’ve been thinking about iPad content, I think as far as AI art goes I covered enough for the iPad itself as the more advanced stuff really is best done on mac.
just discovered this channel and instantly subscribed 😊
Very comprehensive tutorial, thanks!
Both my iPhone and Mac are unable to produce / load any outcomes from prompts. What’s up with that?
This is really great stuff. Id like to introduce it to my high school students but it might be a bit advanced.
FYI, it’s pretty easy to end up in some racy material. The AI crowd likes a lot of questionable material. If you go this route, I’d limit it to the models from Stable Diffusion as they’re responsible, on many fronts. It’s not always clear what models can do, in the SDXL model students could try to generate NSFW but would be severely disappointed.
OMG THANK YOU SO MUCH FOR SHOWING SOME TUTORIALS TOOOO!!! dude the only tutorial about this app i could find on youtube was in Chinese and i don't understand a bit 😭
Try TH-cam thumbnail title like ai art generator to download for iPad specifically because that’s what it’s really for with the pencil there is no other TH-cam video for iPad with SDXL download to iPad for creators-or in my case wanna be creator.
Even tho im a Windows user, i love your vids, are sooo satisfying to watch and also very helpful
(Probably im getting an Intel iMac to use like a secondary PC)
Hey, thanks. Even though I’m a Mac occasionally I find myself watching windows tutorials myself, so I know how it goes.
@@dmug Haha, yeah, i understand
Any tips for a first time Mac experience if i get the iMac?
Get a good GPU, if you are planning to use this app on an intel iMac. Something with a geekbench metal score of 20k or above.
@@tamalchakraborty5346 is an iMac, not the Mac Pro
But just in case, the one im probably getting, has a Radeon 580 Pro, imma do a dual boot to Windows too
Once you go Mac you never go back!
Please make a training video! That would be worth it to me!
Great video, thank you very much.
Thank you for your video, it’s very helpful. Is there a way to add an Url to Stable Diffusion? How do you do it?
Can I use my own artwork
Can I ask it also has plugins like deforum?
It has Loras and Control net
I thought it odd that you used the SDXL refiner model as your base model, instead of the two part base/refiner workflow it was intended for. I hadn’t seen that before!
Its the one that DrawThings vectors you to, it's either a merge or downloads the base in tandem (I can't recall off the top of my head).
Trying to explain checkpoints and such in this video would have been information overload.
How do I set DrawThings back to the default settings?
I don’t have an”generate” button on my iPad version
thanks, great app!
if you explain loras in 15-20 simply it'd be great!
Not a ton to say, just expansion packs to achieve a desired outcome: art style, pose, character, object, setting etc. If you tried to use one designed for photo realism on a model designed to mimic impressionist painting it’s might not work well. It’s a bit of a grab bag. I’d read on AI websites if you want a technical or deep dive, or quiz chat gpt about it.
Hooray for owning a mac. just dove head first into macs 20 years ago! lmao
For some reason I'm getting the first image behind my generated image when I'm following along, and I don't see any tool to get rid of this. It's weird. The zoom with the wheel on the mouse is almost impossible to use because it's so sensitive. I guess there's no way to zoom in without using that.
Tr clicking the new canvas button
@@dmug I'll try that. thanks
Hi. I have been having a play around with Draw things & installed some other AI models. I was just wondering if you know of the "Photoshop" of AI text to image generator tools? As I notice that draw things is unhappy, importing some the bigger (over 4GB) AI models out there and I wondered if their was another tool designed for bigger AI models.
The limitations of larger models is a function of hardware right now. If you're on a lower end or older Mac, then that'll affect it. I've had zero issues with large 12 GB models but I'm also running it primarily on a Mac Pro 2019 with Radeon 6900 XT and M1 Max with 32 GB of RAM.
I'm pretty positive there won't be anything with remotely the polish of Photoshop besides using Photoshop itself with it's own AI tools, at least not for many years. GIMP still feels like a beta of Photoshop 4.0 and that's the best-of for OpenSource image editing. Plus, the technology is young, the method of image generation imprecise, the changes still piling in (see SD1.0 vs SD1.5 vs SD2.0 vs SDXL) and the tools esoteric like poses. The best you'll see is pane-based interfaces like we have right now as that's the case of paid apps that have image generation. They're often only slightly more polished interface wise and much-more featured limited.
This is more difficult then just learning to draw 😂😂😂
Thanks a lot for the explanation of drawthings! Interested in further videos about that.
I am still looking to enhance the resolution of my old low res digital video by IA to hi ret 4K. Nothing work so far.
The best you’ll find is upscaling, using things like Topaz Photo or Pixelmator Pro’s AI upscaler or some of the web utilities.
Unfortunately there’s a real limit to these tools as they can’t infer data that isn’t there so unless you have decent quality reference sources, these tools aren’t super useful. I don’t cover these techniques in this video but I’d try a demo of topaz.
Can I make it learn from my Art ?
Yes but it’s a very deep topic and very slow to perform. Draw things can create LORAs.
It unexpectedly quits for me. Mid 2014 MBP 2.8GHz running the latest Monterey build.
Almost certainly some sort of requirement with metal 3 or such if it’s an iGPU Mac. AI stuff is slow my Mac Pro 2019 and M1 Max
@@dmugMy model has the Nvidia dGPU. I was thinking, is there anything in OCLP that is possibly creating a conflict? I had an issue with OBS/AMFI.
It would be great if you made a Udemy course. There's one about DrawThings AI, but it's really dry, unclear and boring.
Bro, you go that fast. I can barely tell what you’re clicking. You said. Press the camera icon. You did it at this rate which only the gods could see.😂
Yeah I know I zip through this, but I tried to condense the tutorial down to as dense as humanly possible.
what about ai videos
There are stable diffusion video models but the results are pretty bad, you can generate either 14 or 28 frames with the two models. It takes forever, and even using tricks like upscaling, and frame generation in other apps to improve the frame rate, videos are mostly psychedelic smeary messes that are 1-5s long. There’s a few online demos you can play with that are free to use, which do as good of a job (and are faster).
++