@Krypto-dono You know I thought about doing Dall-E 2 as well just not sure if I have an audience for it at this point. Most people are more into SD or MJ. BUT I am still deciding.
@@MonzonMedia Please stick with Stable Diffusion. There are a lot of tutorials on MidJourney already. Not so many for Stable Diffusion. It feels like the majority of AI artists are being left out. Not all AI artists can afford to pay for MidJourney.
I've always ignored the draw image brush for no apparent reason, as I see now that it can be used to create brilliant images. Thank you so much for the inspiration!
Amazed at the many tools available within AI. This was a quick exploration but gives one a taste of some of the tools available. Much to explore and experiment with..
I need to watch this about ten times because I'm pretty sure I am doing things the hard way and making little progress. These videos are a big help. Thank you!
Thank you for the great tutorial. As far as I know I have never seen how this is done, and it explains it easy to understand. You are one of the few who teaches us how to use Playground AI.
You are very welcome and I'm glad to hear that! It's a lot of fun but very addicting hahaha! Be sure to check out this one on building a prompt ua-cam.com/video/r2wJsJEcXZc/v-deo.html.
Loving PAI. Especially the drawing-based image2image tool. Very simple yet allows you to easily place elements and set up a good looking canvas in no time. Having a little in getting very good photorealistic translations though.
Great to hear you are enjoying PAI. In regards to the photorealism, it takes a bit of work to get there from a drawing but very possible. I do it all the time. Anything specific you are having issues with?
@@MonzonMedia No specific issue pertaining to the above. However I do have something I am having a little difficulty with. It seems like if I have a drawn canvas with img2img applied and then try to inpaint an object, the inpainted object often doesn't really adhere to the lighting of the full image. It often has it's own thing going on and looks a little pieced together rather than looking like something that is a part of the image.
@@ish694 It's difficult for me to suggest a solution without actually seeing it however I think I have an idea of what's happening. Part of the limitation in of PAI is that not all the features of stable diffusion are available yet. For example there is mask blur and render in hi-res when you run it locally and it helps with those kinds of results. With that being said, it could be a number of things why that's happening. Changes to the prompt is a typical one. I always tell people try inpainting with the original prompt and add words to the prompt, try not to remove any major words. Also it could be the sampler method, experiment with different ones. Lastly, it also could be the actual inpainting model that PAI is using. There is a default one built into SD 1.5, there is also a separate model just for inpainting when running it locally. I can't say what PAI is using though. If I can recreate that scenario, it might make a good video. 👍
@@MonzonMedia Got it. This is extremely helpful ! Thanks so much for the detailed reply. Would be interesting though if you had a video dissecting the innards of PAI. The models they are using, the settings available, etc.
Hey thank you so much for ur videos they really helped a lot. I was wondering how to use image to image properly on a portrait so that the Style is completly different but keeping the face structure and details so the person can still is to identify Thank you for your help!
I have a video in the works on that topic coming soon. In the mean time you just have to play around with the image strength to get the face where you want it. Mind you it won't be exact but close enough. Then once you have something you like, you can mask the face to change everything else. Less image strength the more random it will be. Hope that helps!
I want to see how i can create an avatar from my own pictures. Or create e very realistic avatar of someone else from their photos. I want to know what promts to use or what filters. Make a video on that please.
Pranta, you've read my mind! 👍The best way is to do training with Dreambooth, or a service like Photo Booth AI then using the ckpt file locally on something like Automatic1111. However, I've played around with a work around that works pretty well just using playgroundai. It's not perfect but you can still be creative. Look out for that to come soon, maybe I'll bump it up from my list of videos to make 😁
Your videos are soooo helpful, Mr. Monzon. Just 2 quick questions. 1st, can you do image-to-image with Easy Diffusion? 2nd, can you do outpainting with Easy Diffusion? Thx.
Hey Michael, I appreciate that! Yes you can do image to image. You just have to hover over the image to get the options and select "use as input" just like you do for inpainting. You'll notice a new slider "prompt strength" the high er it is, the more variation you will get. Then just generate the image. in regards to outpainting, there is plugins available but I'm still looking into it. As of now it doesn't look very easy to install, hoping to find a better way, an easy way. That's why I prefer to use invoke ai since it can do outpainting very well.
May I ask what you are listening on? Laptop, mobile, desktop? Do you have speakers or earphones? Only asking because I'm always looking to improve my audio and I play with my levels a lot 🤭
Unfortunately not possible in Playground but you can use image2image+prompt to blend certain styles. ua-cam.com/video/Bk-FhQsbknM/v-deo.html Might not be exactly what you are looking for but it can work pretty well.
I saw this in a Google Colab but could never get it to work. You provide a Van Gogh image as the style source and your own image as the target. The AI reprocesses your image as if Van Gogh did it himself. Yes, I know you can prompt it but I think a style transfer is less work. Maybe a future addon. Looking forward to your next video and thanks again for the great videos. @@MonzonMedia
@@AnthonyWashingtonX2K Oh I see what you mean! In fact I'm working on a video on this but I might use it for playground ai's channel if they want to use it. If not I will post it on my channel. It's not exactly what you are talking about but a close work around. I do go over it a bit on this video. Sort of a teaser to that. ua-cam.com/video/Bk-FhQsbknM/v-deo.html
You have to see what it is in the image that you don't like. It can be different depending on the image. Some common ones are blurry, over exposed, low quality, over saturated.
Depends on what you are doing, typically I use between 5-15 but mostly for me I find 6-10 is fine for most images. Really just takes some experimenting.
@@MonzonMedia I agree. However, I need to learn the best practice as combining the numbers of prompt guidance, quality and details, seed number, and image strength will take a lot of years to master. The combination from mathematical perspectives is a lot.
Could you show me how to create a consistent character by Playground? Many youtuber did it with Mid Journey, but I can't do the same thing with Playground
It's a bit tricky without being able to use Dreambooth or a service like Open AI's Photobooth, however I do have a work around hack that works pretty well for generating the same face through prompts only. I will post hopefully tomorrow.
You're welcome! I haven't heard personally but I'm hoping they will at least get outpainting out soon Thing is for something like that, you need some serious GPU power so that could be quite costly. I'll see if I can find out anything. 👍
Yes, absolutely. I've been getting familiar for a couple months now. Recently they launched a way to install models fairly easily so now I can create some content on Invoke AI. It's outpainting feature is probably the best I've seen so far. Are you using it?
@@MonzonMedia no not yet. I'm pretty new. I just did some testing on DALLE 2, and more on playground AI. But this week I will use it. Do you plan to do video on it? Thanks again for your work!
The ramp of image strength should probably be turned down to be honest. It's an AI in there going "omg I changed every Y value of every pixel by 3, but here's the kicker 3 UP OR 3 DOWN IT WAS RANDOM" and humans look at it and can't see that any change was made. But yeah anyway midjourney can get fucked with that pricing model and drugged out marketing. And the org itself is a goddam mess. Using this one has been ***wonnderfullll***
hahaha yeah it can get out of hand if it's not used properly. Glad you are enjoying the playground content, in case you don't know most of my content on playground ai will be on the main channel. I'll still cover some things here but most of it will be on the official channel. Appreciate the support! www.youtube.com/@playgroundai
Playground is AI is hit and miss I loaded up the app on a very expensive phone and it wouldn't generate any image for me and when I went and looked at other people's complaints the complaints of other people are endless endless endless is the same thing is happening to them with their phones they opened up an account and it just keeps using up the points and kicking out the message failed to generate failed to generate failed to generate who wants to deal with a program like that that is so hit and miss that is pure outlaw
Please stick with Stable Diffusion. There are a lot of tutorials on MidJourney already. Not so many for Stable Diffusion. It feels like the majority of AI artists are being left out. Not all AI artists can afford to pay for MidJourney.
I've been trying for 2 days, but I can't place an Anjou lily or coat of arms on the body of a figure or on any image. Can you give me an idea? For example, how could this motif be merged from one picture to another? Have any idea?
I would try to use the inpaint function first, just mask the area and enter Anjou lily in your prompt. You can also try to do it in the new outpaint feature, works the same way but more flexibility. ua-cam.com/video/6RFdhduGf9A/v-deo.html
I've got a more in-depth tutorial on image to image coming soon. I also want to cover some Invoke AI Outpainting. What else would you like to see?
@Krypto-dono You know I thought about doing Dall-E 2 as well just not sure if I have an audience for it at this point. Most people are more into SD or MJ. BUT I am still deciding.
@@MonzonMedia Please stick with Stable Diffusion. There are a lot of tutorials on MidJourney already. Not so many for Stable Diffusion. It feels like the majority of AI artists are being left out. Not all AI artists can afford to pay for MidJourney.
I've always ignored the draw image brush for no apparent reason, as I see now that it can be used to create brilliant images. Thank you so much for the inspiration!
Now you know! It’s actually pretty amazing what you can do with simple line art 😁 glad it was helpful and really appreciate the feedback!
Amazed at the many tools available within AI. This was a quick exploration but gives one a taste of some of the tools available. Much to explore and experiment with..
Yes and it keeps changing. Even better for us that gets to discover all these tools. Thanks for checking out the vid! 🙌
Your videos are very easy to follow yet not dragging. Thanks a lot!!!
I appreciate that positive feedback my friend, I'm always trying to improve on my delivery. 👍
I need to watch this about ten times because I'm pretty sure I am doing things the hard way and making little progress. These videos are a big help. Thank you!
You got this! Is there anything in particular you are struggling with?
Thank you for the great tutorial. As far as I know I have never seen how this is done, and it explains it easy to understand. You are one of the few who teaches us how to use Playground AI.
I really appreciate that Richard and glad to hear it was helpful. I’ll have more on this topic soon! Have a great weekend!
Thank you so much for the prompts. This is my first time I am getting this professional looking images in SD.
You are very welcome and I'm glad to hear that! It's a lot of fun but very addicting hahaha! Be sure to check out this one on building a prompt ua-cam.com/video/r2wJsJEcXZc/v-deo.html.
@@MonzonMedia Sure, i will check them.
Great SD tutorial. Clear and concise. Well paced and easy to follow along. Subscribed. Thank you. 💯👍
Appreciate the support! You’re welcome! 🙌🏼👍🏼
thank you so much
You are very welcome! Be sure to check out Playground's official channel ---> www.youtube.com/@playgroundai
Thanks 👍
Welcome 👍
Useful little demo, thanks.
Glad to hear it was useful for ya! Thanks for checking it out! 🙌
Loving PAI. Especially the drawing-based image2image tool. Very simple yet allows you to easily place elements and set up a good looking canvas in no time. Having a little in getting very good photorealistic translations though.
Great to hear you are enjoying PAI. In regards to the photorealism, it takes a bit of work to get there from a drawing but very possible. I do it all the time. Anything specific you are having issues with?
@@MonzonMedia No specific issue pertaining to the above. However I do have something I am having a little difficulty with.
It seems like if I have a drawn canvas with img2img applied and then try to inpaint an object, the inpainted object often doesn't really adhere to the lighting of the full image. It often has it's own thing going on and looks a little pieced together rather than looking like something that is a part of the image.
@@ish694 It's difficult for me to suggest a solution without actually seeing it however I think I have an idea of what's happening. Part of the limitation in of PAI is that not all the features of stable diffusion are available yet. For example there is mask blur and render in hi-res when you run it locally and it helps with those kinds of results. With that being said, it could be a number of things why that's happening. Changes to the prompt is a typical one. I always tell people try inpainting with the original prompt and add words to the prompt, try not to remove any major words. Also it could be the sampler method, experiment with different ones. Lastly, it also could be the actual inpainting model that PAI is using. There is a default one built into SD 1.5, there is also a separate model just for inpainting when running it locally. I can't say what PAI is using though. If I can recreate that scenario, it might make a good video. 👍
@@MonzonMedia Got it. This is extremely helpful ! Thanks so much for the detailed reply. Would be interesting though if you had a video dissecting the innards of PAI. The models they are using, the settings available, etc.
Great tutorial, thank you.
You are welcome and my pleasure! Thanks for taking the time to watch and comment. 👍
Hey thank you so much for ur videos they really helped a lot. I was wondering how to use image to image properly on a portrait so that the Style is completly different but keeping the face structure and details so the person can still is to identify
Thank you for your help!
I have a video in the works on that topic coming soon. In the mean time you just have to play around with the image strength to get the face where you want it. Mind you it won't be exact but close enough. Then once you have something you like, you can mask the face to change everything else. Less image strength the more random it will be. Hope that helps!
I should try turning my sketches into masterpieces too. 😁
Sure why not! Very fun to see how you can stretch your creativity
I want to see how i can create an avatar from my own pictures. Or create e very realistic avatar of someone else from their photos. I want to know what promts to use or what filters. Make a video on that please.
Pranta, you've read my mind! 👍The best way is to do training with Dreambooth, or a service like Photo Booth AI then using the ckpt file locally on something like Automatic1111. However, I've played around with a work around that works pretty well just using playgroundai. It's not perfect but you can still be creative. Look out for that to come soon, maybe I'll bump it up from my list of videos to make 😁
Your videos are soooo helpful, Mr. Monzon. Just 2 quick questions. 1st, can you do image-to-image with Easy Diffusion? 2nd, can you do outpainting with Easy Diffusion? Thx.
Hey Michael, I appreciate that! Yes you can do image to image. You just have to hover over the image to get the options and select "use as input" just like you do for inpainting. You'll notice a new slider "prompt strength" the high er it is, the more variation you will get. Then just generate the image. in regards to outpainting, there is plugins available but I'm still looking into it. As of now it doesn't look very easy to install, hoping to find a better way, an easy way. That's why I prefer to use invoke ai since it can do outpainting very well.
I still struggle with mask and inpainting - they at times seem to do the same
Technically they are one in the same, isolating an area to focus on. 👍
pls bring your volume up tia!
May I ask what you are listening on? Laptop, mobile, desktop? Do you have speakers or earphones? Only asking because I'm always looking to improve my audio and I play with my levels a lot 🤭
@@MonzonMediadefinitely! Laptop with earbuds. Just a tad bit louder would be good.
@@strangerdaysss Thank you for the feedback. Other videos are a bit louder, or at least should be :)
how to add a background to my image like place her in a scene of different kinds
You can use inpainting to mask the background and modify the prompt accordingly.
how to blend to face in playground ai just like mid journey??? /blend????
Unfortunately not possible in Playground but you can use image2image+prompt to blend certain styles. ua-cam.com/video/Bk-FhQsbknM/v-deo.html Might not be exactly what you are looking for but it can work pretty well.
how do i make somthing into flower art?
How do you save your images?
Thanks for the video, lots of great info.
Is there a way to do a style transfer in Playground AI?
You are very welcome! I'm not sure what you mean by style transfer?
I saw this in a Google Colab but could never get it to work. You provide a Van Gogh image as the style source and your own image as the target. The AI reprocesses your image as if Van Gogh did it himself. Yes, I know you can prompt it but I think a style transfer is less work. Maybe a future addon. Looking forward to your next video and thanks again for the great videos. @@MonzonMedia
@@AnthonyWashingtonX2K Oh I see what you mean! In fact I'm working on a video on this but I might use it for playground ai's channel if they want to use it. If not I will post it on my channel. It's not exactly what you are talking about but a close work around. I do go over it a bit on this video. Sort of a teaser to that. ua-cam.com/video/Bk-FhQsbknM/v-deo.html
What are the negative prompts for nature?
You have to see what it is in the image that you don't like. It can be different depending on the image. Some common ones are blurry, over exposed, low quality, over saturated.
@@MonzonMedia How to deal with rainbow that keeps appearing on the ground? Or a bird that seems to stand in a tree that does not exist?
What is the optimal number of Prompt Guidance?
Depends on what you are doing, typically I use between 5-15 but mostly for me I find 6-10 is fine for most images. Really just takes some experimenting.
@@MonzonMedia I agree. However, I need to learn the best practice as combining the numbers of prompt guidance, quality and details, seed number, and image strength will take a lot of years to master. The combination from mathematical perspectives is a lot.
Cool video can you make a video on how to have high quality images for print? if you know how of course thank you😊
Definitely something I've been meaning to cover. Appreciate you watching and for the feedback!
Could you show me how to create a consistent character by Playground? Many youtuber did it with Mid Journey, but I can't do the same thing with Playground
It's a bit tricky without being able to use Dreambooth or a service like Open AI's Photobooth, however I do have a work around hack that works pretty well for generating the same face through prompts only. I will post hopefully tomorrow.
@@MonzonMedia
Hello, thanks again for your video.
Do you know if "generation frame is planed" on playground?
You're welcome! I haven't heard personally but I'm hoping they will at least get outpainting out soon Thing is for something like that, you need some serious GPU power so that could be quite costly. I'll see if I can find out anything. 👍
@@MonzonMedia thanks!
Playground AI only draws from the front view. Can drawings be made from other perspectives?
Of course, you just have to prompt for it, for example "side view portrait of a man"
Is this the free version 😊
At the time of recording this video yes. 👌
@@MonzonMedia Wow
@@MonzonMedia If I have a bunch of scenery photos, can I turn them into paintings and so on?
Did you ever use invoke AI?
Yes, absolutely. I've been getting familiar for a couple months now. Recently they launched a way to install models fairly easily so now I can create some content on Invoke AI. It's outpainting feature is probably the best I've seen so far. Are you using it?
@@MonzonMedia no not yet. I'm pretty new. I just did some testing on DALLE 2, and more on playground AI.
But this week I will use it. Do you plan to do video on it?
Thanks again for your work!
Yes indeed! Keep an eye out for it. 👍
The ramp of image strength should probably be turned down to be honest. It's an AI in there going "omg I changed every Y value of every pixel by 3, but here's the kicker 3 UP OR 3 DOWN IT WAS RANDOM" and humans look at it and can't see that any change was made.
But yeah anyway midjourney can get fucked with that pricing model and drugged out marketing. And the org itself is a goddam mess. Using this one has been ***wonnderfullll***
hahaha yeah it can get out of hand if it's not used properly. Glad you are enjoying the playground content, in case you don't know most of my content on playground ai will be on the main channel. I'll still cover some things here but most of it will be on the official channel. Appreciate the support! www.youtube.com/@playgroundai
Playground is AI is hit and miss I loaded up the app on a very expensive phone and it wouldn't generate any image for me and when I went and looked at other people's complaints the complaints of other people are endless endless endless is the same thing is happening to them with their phones they opened up an account and it just keeps using up the points and kicking out the message failed to generate failed to generate failed to generate who wants to deal with a program like that that is so hit and miss that is pure outlaw
Please stick with Stable Diffusion. There are a lot of tutorials on MidJourney already. Not so many for Stable Diffusion. It feels like the majority of AI artists are being left out. Not all AI artists can afford to pay for MidJourney.
Appreciate the feedback. I do tend to cover more SD topics so no worries there. 👍🏼
@@MonzonMedia Glad to hear this !
I've been trying for 2 days, but I can't place an Anjou lily or coat of arms on the body of a figure or on any image. Can you give me an idea? For example, how could this motif be merged from one picture to another? Have any idea?
I would try to use the inpaint function first, just mask the area and enter Anjou lily in your prompt. You can also try to do it in the new outpaint feature, works the same way but more flexibility. ua-cam.com/video/6RFdhduGf9A/v-deo.html