yeah playgroundAI is definetly a huge W, it has simple GUI, a whopping *1000* images per day, and just makes things super convenient. all that for free too.
@@MattVidPro this particular system might even be complimented by having the damned thing running on an external GPU housing. Something on the order of Puget Sound.
MidJourney makes consistently beautiful, detailed and compelling images. It's like having an extremely talented artist working for you. It's not easy to get specific details or composition with it without knowing a lot of advanced techniques. In terms of quality, MidJourney is still producing the best looking images.
I am using playgroundai now for my visual novel. You can change facial expression through inpainting. The fact that you can generate so many images for free is a huge plus. Even after 1000 images in one day you can still create images for free, because it only adds a waiting time of about 1 minute for a new picture. What I would like to learn is how to 'train a model or avatar', so that you can make it maintain for example same clothing and hairstyle through different poses and settings. I saw some guide for that in Midjourney but not yet in playgroundai.
Did you know an A.I. helped me on my 17min animation? But I released the copyright to the public domain just because I didn't plan on monetizing it with ads. Plus, I don't know which artists they trained the diffusions on. I think that A.I. art alone by itself isn't respected--and that's why many artists hate it. But if we turn it into something meaningful, like an informational tool, then we are adding value to it; becomes a new experience.
Here's a killer trick with playground. Download the upscaled version of a flawed image you've created. Then, plug that into the image to image slot and use their mask tool to blot out the objectionable portion and boom. Run those at high resolution, of course. Use the original prompt and pare back the parts that don't apply to the thing you're fixing. You very effectively fix bad hands, faces, etc. I haven't tried it, but I bet you could add characters this way, too. I've definitely used it to add silly things to pre-existing images.
I found out about Playground thanks to your video that u initially did on it and have been using it ever since. I also use MJ everyday. In my experience MJ is still ahead in its overall capabilities of the aesthetics it can produce. However I would say that if you are patient and start mastering negative prompts in Playground, you can definitely produce amazing imagery. To date I would say that the Image to Image feature + Negative prompts without any filters have created some mind blowing imagery on Playground for me, but as I said it really does take some patience. Thanks for another cool video amigo 👍🏽
They made their own Stable Diffusion fine-tuned model, which is cool. I dig their web ui Locally generating with Analog Diffusion FTW, IMHO. Midjourney seems to be pre-stuffed with Analog style prompts. They pad prompts. That would be funny for you to show in a video: Just type gibberish into MJ such as “xyzcat” and you’ll still get a highly detailed analog photo style image generated.
i am COMPLETELY new to AI in every way, but i landed on playground a week ago and cannot stop! ive seen some great stuff in MJ4 but i dont want to pay that much until i see if i am going to keep this up, so far im very pleased with playground. thanks a lot for this video!
This is a wonderful episode, I have learned a lot of prompts technuques, very useful and inspirational, this channel is a gem especially for people who are not familiar with arts but want to generate stunning images, thank you so much 🙏
Playground AI is using really powerful GPU's. I'm running the same settings, same sampler, resolution, stable diffusion, etc on my GTX 1070 and it's obviously much slower for me. Playground's comes out really fast.
I've been using playground for a couple of days and the images I get and or edit are really good but its way too easy to trip the dreaded "Error: Your request activated the API's safety filters and could not be processed". I'm using what I would image to be safe prompts without anything that includes violence or anything sexual in nature and still get this error from time to time. I really like the results and can only see things getting better but they need to relax the safety filters a bit. Hands are still an issue but I am doing mostly portrait studio head shots. MJ has been really great as well - great vid
Great video. I am particularly impressed by the audio quality. Could you comment on how it was done ? Was it post processed. Very good articulation as well. Obviously, many thanks for the content as well
A problem with ChatGPT is that it is temporary. I am curious to see how openai will manage its end. They will probably make it a paid service but that will frustrate many.
@@CypherDND Short answer : no. Openai seems take a great care not to open release its models. It can be because they want to keep control either of more or less malicious use ou to better fit a business model.
Why would you pay for this when you can do it on your own PC for free? More of a learning curve but waaaay more control and options as well as zero restrictions on what you create.
@@buttson707 it certainly helps, and there are some specific things that would require it. But for general use plenty of people are doing it with older 10xx series gpu's. It is primarily dependent on how much vram you have from my understanding.
@@MattVidPro that's understandable, makes complete sense. As for dreambooth, it's working in automatic1111, and I'm pretty sure you can make whatever you like with it. Understandable why people might not want to learn how to do it on their own, but a constant stream of user created ones are released all the time.
I would like to thank you so much for your jarfix tutorial. If I never saw that video I think I would've just quit Minecraft. Thank you so so so much!!!
@@fnorgenI have found doing it locally gets me waaaay better results with a lot more accuracy. A learning curve for sure, but the options and tweaks are almost limitless. There are some very good models on civitai, minimum prompts with very good results in invoke or automatic1111
i have a serious question tho can smone help ?? i been using this for like 4 days now and when i was about to upscale an image recently it said that i reached the 30 image free upscale a day soo what does this mean ?? i always upscaled freely and only happened today, will i be able to upscale again after 24h or what ?? also what will happen exactly after i reach 1000 generated pics ??
Does anyone know if Playground AI pro version high resolution option actually produces better added details vs their upscale 4x which is just upscaling the image from a low res one without adding any actual zoom in details which turns out mediocre in quality?
Just a thought about AI Art being theft. Don't human artists get inspiration from other artists' work? If they draw something similar in style, is that considered theft?
Hi. The Google Photo print service is apparently blocking Midjourney prints. I am unable to order a canvas print. I can do any other prints other than a MidJourney print. Has anyone else had this issue? What print services do you use?
Yes, hi...I have been using both befiore I stumbled upon your thread and gladly subscribed to your RUclips and Discord but, about these 2, Playground AI and Midjourney....I use my OWN imagery that I have drawn and then, used the AI Generator, the ONLY AI Generator I see that ends up with something Satisfactory is Midjourney....I have about 9 different AI Art Gen and Stable Diffusion Generators and nothing really Finished with anything even remotely useful...Maybe you can help! Thank you!
The one thing about playground ai is that it never gives me one person image it be having like 4 heads in one photo I don’t know how to make just one person in a person
i dont like how midjourney is discord only and all the demos ive seen of it look generic, like how many times can somebody generate "generic pretty woman" and show it everywhere like its the greatest thing ever
I agree the examples are very boring but if u look up creators on instagram for examples, they get really creative with it, I follow someone that generates non existent movie scenes with a specific vintage style which I love, nothing generic and soulless. Really comes down to the creators taste and aesthetic
Well, I agree with you about art, but my favorite is chatGPT as an assistant. Thanks to you and others AI has become one of my main interests...I'm starting to learn PyTorch. Here is an Ode To A Lonely Tensor: A lonely tensor sat, In a matrix all alone. No other numbers by its side, No friends to call its own. It multiplied and added, But no one was around To share its joy or sorrow, No comforting sound. But one day, a vector came And joined the lonely tensor's game. Together they solved equations, And the tensor was no longer the same. It had found a friend, Someone to share its space. Together they were stronger, A perfect pair in this numerical place. So here's to the lonely tensor, No longer on its own. With a friend by its side, It's found a happy home.
This link do not work, from link from videos: ruclips.net/user/MattVideoProductions This link works from Subscription list: www.youtube.com/@MattVidPro
Finally a great option for web image generation for free (that I know of). Since lexica became paid I was wondering where I would play with this stuff next
Does PlaygroundAI have a private mode, or will our generations be public for everyone to see? If yes, is the private mode in PlaygroundAI also free? I’m not a fan of Midjourney because of the lack of privacy. I also don’t like that Midjourney makes their users pay extra just to keep their generations private. I wonder if PlaygroundAI is a good alternative to Midjourney.
What's the cheapest way to make a large amount of avatars with dream booth as efficiently as possible, without having to use platforms such as stable diffusion on gradio as I'm struggling with prompts?
Playground is good but it's not yet on Midjourney level, it's great for certain styles but for anime or cartoon it's very bad. The hands and feet at also systematically bad and when you want something else other than a portrait playground is struggling.
Share your Art in my DISCORD! discord.gg/bQgcbjs2Sg
yeah playgroundAI is definetly a huge W, it has simple GUI, a whopping *1000* images per day, and just makes things super convenient. all that for free too.
Absolutely!! Its a huge W
@@MattVidPro W chain
I just wish I could get decent results with it. Everything I’ve generated so far has been just horrendous.
@@tarabooartarmy3654 yeah its still stable diffusion so its quite limited
@@MattVidPro this particular system might even be complimented by having the damned thing running on an external GPU housing. Something on the order of Puget Sound.
MidJourney makes consistently beautiful, detailed and compelling images. It's like having an extremely talented artist working for you. It's not easy to get specific details or composition with it without knowing a lot of advanced techniques. In terms of quality, MidJourney is still producing the best looking images.
I am using playgroundai now for my visual novel. You can change facial expression through inpainting. The fact that you can generate so many images for free is a huge plus. Even after 1000 images in one day you can still create images for free, because it only adds a waiting time of about 1 minute for a new picture.
What I would like to learn is how to 'train a model or avatar', so that you can make it maintain for example same clothing and hairstyle through different poses and settings. I saw some guide for that in Midjourney but not yet in playgroundai.
By accident instead of a realistic I got a child’s book like lion - tada now I can make my kids book stories in no time
SD and MidJourney have surpassed Dalle 2 IMO. Hopefully OpenAI will make Dalle 2 Open Source.
Most definitely! this proves it!
yeah by a long shot
I feel they won't though until they're lapped even further.
In everything except outpainting. Dalle 2 is still the king of outpainting.
@@dissonancE.. true, dall-e 2s coherency for images is very good
Did you know an A.I. helped me on my 17min animation?
But I released the copyright to the public domain just because I didn't plan on monetizing it with ads.
Plus, I don't know which artists they trained the diffusions on.
I think that A.I. art alone by itself isn't respected--and that's why many artists hate it.
But if we turn it into something meaningful, like an informational tool, then we are adding value to it; becomes a new experience.
Here's a killer trick with playground. Download the upscaled version of a flawed image you've created. Then, plug that into the image to image slot and use their mask tool to blot out the objectionable portion and boom. Run those at high resolution, of course. Use the original prompt and pare back the parts that don't apply to the thing you're fixing. You very effectively fix bad hands, faces, etc. I haven't tried it, but I bet you could add characters this way, too. I've definitely used it to add silly things to pre-existing images.
What do you mean?
It's good that you always leave links 🤩
I found out about Playground thanks to your video that u initially did on it and have been using it ever since. I also use MJ everyday. In my experience MJ is still ahead in its overall capabilities of the aesthetics it can produce. However I would say that if you are patient and start mastering negative prompts in Playground, you can definitely produce amazing imagery. To date I would say that the Image to Image feature + Negative prompts without any filters have created some mind blowing imagery on Playground for me, but as I said it really does take some patience. Thanks for another cool video amigo 👍🏽
They made their own Stable Diffusion fine-tuned model, which is cool. I dig their web ui
Locally generating with Analog Diffusion FTW, IMHO.
Midjourney seems to be pre-stuffed with Analog style prompts. They pad prompts.
That would be funny for you to show in a video: Just type gibberish into MJ such as “xyzcat” and you’ll still get a highly detailed analog photo style image generated.
Agree, a local install of Analog Diffusion has had me distracted for a week
@@beardordie5308 right?! Not just with people/portraits but with landscapes/products/cars etc etc
i am COMPLETELY new to AI in every way, but i landed on playground a week ago and cannot stop! ive seen some great stuff in MJ4 but i dont want to pay that much until i see if i am going to keep this up, so far im very pleased with playground. thanks a lot for this video!
Welcome aboard!
You was hovering over my “happy monster”😅. I watch every video you release, so its cool to see Playground review again!🥳
Sweet generation! Thank you for watchingg!!
It can actually generate new knowledge:
New architecture
New product designs
New steampunk inventions
11:58 ... thats an interesting finger shape there lol
thanks i cant unsee it
I made the middle pic at 3:19. I've been using playgroundai since I saw your video a few months ago.
Sweet gen! Thanks for watching :)
This is a wonderful episode, I have learned a lot of prompts technuques, very useful and inspirational, this channel is a gem especially for people who are not familiar with arts but want to generate stunning images, thank you so much 🙏
Many thanks!
Playground AI has cured me of my Twitter addiction. Fortunately I now have a Playground AI addiction. X-D
Haha so true
one tip that might help, adding "anthropomorphic" to your prompt
Very true, never thought to do that
Playground AI is using really powerful GPU's. I'm running the same settings, same sampler, resolution, stable diffusion, etc on my GTX 1070 and it's obviously much slower for me. Playground's comes out really fast.
I've been using playground for a couple of days and the images I get and or edit are really good but its way too easy to trip the dreaded "Error: Your request activated the API's safety filters and could not be processed". I'm using what I would image to be safe prompts without anything that includes violence or anything sexual in nature and still get this error from time to time. I really like the results and can only see things getting better but they need to relax the safety filters a bit. Hands are still an issue but I am doing mostly portrait studio head shots. MJ has been really great as well - great vid
AI art is amazing now. No stopping it. I can make anything I want.
I use Playgroundai because it’s much easier and direct than MJ. A lot of fun with image to image function.
Appreciate your updates as ever, Matt. So, can we load these embeddings into a local Stable Diffusion install?
No, you can only use them for playground, thanks for watching
Thanks for pointing me to playground!
This is great for someone who can't get what their thinking in their head to paper. I always redo what I get
Great video. I am particularly impressed by the audio quality. Could you comment on how it was done ? Was it post processed. Very good articulation as well. Obviously, many thanks for the content as well
Thank you so much for mentioning that site!
Thats what Im here for!
I’d still use MidJourney over that just cuz of the quality and the amount of stuff u could work with
Hey matt, you should look into "Karlo AI" if you haven't already!
It there an IA that can generate images, but then to which we can modifiy the images generated by writing what changes we want in it?
Love your channel broski :)
keep it up!!!
Glad you enjoy it!
Are the images taken on playground copyright?
Happy New Year!
Tell me, is there anything for AMD? Are they working in this direction?
Why doesn't SD work on AMD?
Wow 1000 free images a day is insane! Curious if decent logos can be made (and does this have the issues w text that MidJourney does)
Please can you do more AI roguelite videos? It got a new update with Text to Speech features!
ooh yes. I enjoy playground Ai. Thank you for bringing more enlightenment to this website
A problem with ChatGPT is that it is temporary. I am curious to see how openai will manage its end. They will probably make it a paid service but that will frustrate many.
Is there any alternative to ChatGPT that can be run locally on your own machine?
@@CypherDND Short answer : no.
Openai seems take a great care not to open release its models. It can be because they want to keep control either of more or less malicious use ou to better fit a business model.
this is stable diffusion 2.1?
You can generate with DAll-E 2 on there as well, just like Night Cafe
Yes!
And it's cheaper than Dall e 2 official website bcoz playground gets bulk discount on api
Looks like the added traffic broke playground. I've tried 4 times and always get a 500 (internal service error).
People are reporting that… terribly sorry to hear I’m sure it will be fixed soon
Honestly I just canceled my subscription on Midjurney and started using playground 😀
Nahh bruh why were all the lemon characters at the end actual nightmare fuel 💀💀
😂😂😂
Nice video!!!
thank you
Been using PlaygroundAi for a few months.
Why would you pay for this when you can do it on your own PC for free? More of a learning curve but waaaay more control and options as well as zero restrictions on what you create.
doesn't it need a super beefy PC to generate these with good speed
@@buttson707 it certainly helps, and there are some specific things that would require it. But for general use plenty of people are doing it with older 10xx series gpu's. It is primarily dependent on how much vram you have from my understanding.
Most dont have the computer set up, and these dreambooth filters cannot be run locally
@@MattVidPro that's understandable, makes complete sense. As for dreambooth, it's working in automatic1111, and I'm pretty sure you can make whatever you like with it. Understandable why people might not want to learn how to do it on their own, but a constant stream of user created ones are released all the time.
I used Playground, but ...my biggest problem is the lack of making panoramas style, portrait look is kind'a borring
Wow this video was very informative
I would like to thank you so much for your jarfix tutorial. If I never saw that video I think I would've just quit Minecraft. Thank you so so so much!!!
Can you please test the img to img function?
Can you make a video on the codes used to create images on any AI Platform
feel like ai music is gonna take over soon
it might already have....
But can you input your own image into playground ie charater desingn and celebrities
Yes you can
Can you use these filters on images generated on other platforms? Could I use face restoration on a Midjourney-made image?
Midjourney is far better in term of result, anatomy, perspective, colours, simple prompts and doing other things that just portraits...
I still prefer using stable diffusion on my own machine, with all features. This just look like a "demo" version to me
Not local -> not interested.
Still though, I'm a little envious at how good some of these secret sauce online platforms are.
@@fnorgenI have found doing it locally gets me waaaay better results with a lot more accuracy. A learning curve for sure, but the options and tweaks are almost limitless. There are some very good models on civitai, minimum prompts with very good results in invoke or automatic1111
Not everyone has a computer with a modern graphics card. I personally use these web applications with my phone
@@victorsantos9884 That makes sense.
i have a serious question tho can smone help ?? i been using this for like 4 days now and when i was about to upscale an image recently it said that i reached the 30 image free upscale a day soo what does this mean ?? i always upscaled freely and only happened today, will i be able to upscale again after 24h or what ?? also what will happen exactly after i reach 1000 generated pics ??
HI we can use the photos of this site to design t-shirts ,is there any copyright problem?
LMAO I just tried to edit a photo of a cowboy metal band playing live and ended up with green grass. :P
Congratulations for the video, so this site can be created and downloaded for free?
Thanks, I subbed.
It only has 2 filters now?
After 3 minutes into the video I could only abort to watch it. Cuz I had to try it out immediately myself! And it's 2 hours later now... xD
Are the filters dreambooth models or hypernetworks?
I believe they are dreambooth models
Does anyone know if Playground AI pro version high resolution option actually produces better added details vs their upscale 4x which is just upscaling the image from a low res one without adding any actual zoom in details which turns out mediocre in quality?
Just a thought about AI Art being theft. Don't human artists get inspiration from other artists' work? If they draw something similar in style, is that considered theft?
Hi. The Google Photo print service is apparently blocking Midjourney prints. I am unable to order a canvas print. I can do any other prints other than a MidJourney print.
Has anyone else had this issue? What print services do you use?
how would they know and why would they care ?
just use any other Printing service or well... buy a good printer if it hasnt to be Canvas.
I will be able to remedy this with a video soon but for now I would use printful
Yes, hi...I have been using both befiore I stumbled upon your thread and gladly subscribed to your RUclips and Discord but, about these 2, Playground AI and Midjourney....I use my OWN imagery that I have drawn and then, used the AI Generator, the ONLY AI Generator I see that ends up with something Satisfactory is Midjourney....I have about 9 different AI Art Gen and Stable Diffusion Generators and nothing really Finished with anything even remotely useful...Maybe you can help! Thank you!
Midjourney has the best image to image. Especially with multiple source images you want to mix.
Yeah img to img is a growing subset of image AI and midjourney does it the best at the moment
Great video.
The one thing about playground ai is that it never gives me one person image it be having like 4 heads in one photo I don’t know how to make just one person in a person
Playground AI site won't load for me. Is it down?
Can you import a reference image/photo ?
i dont like how midjourney is discord only and all the demos ive seen of it look generic, like how many times can somebody generate "generic pretty woman" and show it everywhere like its the greatest thing ever
good point, I wish you could gen on their website
thank u someone who understands
Agreed yah
Most of the images on playground AI main feed are different variations of women. That's boring.
I agree the examples are very boring but if u look up creators on instagram for examples, they get really creative with it, I follow someone that generates non existent movie scenes with a specific vintage style which I love, nothing generic and soulless. Really comes down to the creators taste and aesthetic
Well, I agree with you about art, but my favorite is chatGPT as an assistant. Thanks to you and others AI has become one of my main interests...I'm starting to learn PyTorch. Here is an
Ode To A Lonely Tensor:
A lonely tensor sat,
In a matrix all alone.
No other numbers by its side,
No friends to call its own.
It multiplied and added,
But no one was around
To share its joy or sorrow,
No comforting sound.
But one day, a vector came
And joined the lonely tensor's game.
Together they solved equations,
And the tensor was no longer the same.
It had found a friend,
Someone to share its space.
Together they were stronger,
A perfect pair in this numerical place.
So here's to the lonely tensor,
No longer on its own.
With a friend by its side,
It's found a happy home.
excellent. a tool for graphic designers.
Absolutely!
Playground generates creepy image
Some reason I can't open your channel main page..?
This link do not work, from link from videos:
ruclips.net/user/MattVideoProductions
This link works from Subscription list:
www.youtube.com/@MattVidPro
Playground AI - cool!!!!!!!!!!!
i haven't been able to make anything even close to the showcase with Stable Diffusion Seems scam :(
Finally a great option for web image generation for free (that I know of). Since lexica became paid I was wondering where I would play with this stuff next
Does PlaygroundAI have a private mode, or will our generations be public for everyone to see? If yes, is the private mode in PlaygroundAI also free?
I’m not a fan of Midjourney because of the lack of privacy. I also don’t like that Midjourney makes their users pay extra just to keep their generations private. I wonder if PlaygroundAI is a good alternative to Midjourney.
They do have a private mode which I believe is the default. Can later chose to let the image go public if you so choose.
Wot if u delete your images in midjourney? I deleted a bunch and they are gone.
Privacy = delete (red x emoji)
Why do people make the same oversaturated face portraits?
and its always the Same type of Woman that looks like a Instagram Filter.
@@mordisgaminggedons5119 exactly. At least make something that looks interesting
Without a tree, you won't make juice
What's the cheapest way to make a large amount of avatars with dream booth as efficiently as possible, without having to use platforms such as stable diffusion on gradio as I'm struggling with prompts?
tough call... the discord server would probably know I would try asking the forums: discord.gg/bQgcbjs2Sg
Why can't I go to your channel? It says 404 page isnt available?
RUclips screwed up links with the launch of handles
I don’t suppose you ever had a lemonade stand as a kid? Lol
Not where I live LOL
Hi, please enable subtitles
This, every video should have CC
What is it with you and lemons?!
What is the discord?
Link in desc
I just tried it. I was disappointed. Unless I didn’t know what I was doing 😢
Playground is good but it's not yet on Midjourney level, it's great for certain styles but for anime or cartoon it's very bad.
The hands and feet at also systematically bad and when you want something else other than a portrait playground is struggling.
epic matt video
Thanks!
Mid journey has a pay wall and so yes I like free
Ye
Free plan limits quality and detail after 50 images !
Maybe try incognito mode?
Doesnt look better than Midjpurney, but the price point is a win
Midjourney is the best!
I think you oversold playground AI
Niji journey is amazing, better than midjourney v4 in my experience
👏👏👏
Yes Al art is always a good topic .
alot of people also believe the Earth is flat.