Perfect Prompts Made Easy - Midjourney & PlaygroundAI
Вставка
- Опубліковано 18 жов 2024
- In this video, we will show you how to create realistic AI art prompts using ChatGPT. We will begin by discussing the different mediums that can be used, such as digital art, oil paintings, watercolors, sketches, and photography.
Next, we will cover how to describe the person or subject in the artwork, including their age, hair color, hairstyle, hair length, eye color, eye shape, facial expression, and clothing. It is important to be as detailed as possible in order to get the most realistic results.
We will then move on to describing the background of the artwork, including the environment, lighting, and any additional objects or decorations that should be included. The more specific and detailed the description, the more realistic the resulting image will be.
We will also discuss how to direct the pose of the person or subject in the artwork, including the position of their head, body, and limbs, as well as any facial expressions or gestures. This will help create a dynamic and interesting image.
Finally, we will show you how to use artist names as style references. This is a great way to emulate the style of a particular artist, and can help to create a unique and interesting image.
By following these tips and guidelines, you will be able to create realistic AI art prompts that will produce stunning and detailed images. Whether you are creating digital art, oil paintings, sketches, or photographs, you can use these prompts to produce high-quality, realistic artwork. So why wait? Give it a try and see the magic of AI art creation unfold right before your eyes!
Email: john@aijohn.art
Twitter: / aijohnart
Instagram: / aijohnart
As promised, here is the text I entered into ChatGPT as well as a link for PlaygroundAI.com
"I have an AI that outputs an image when I enter a prompt. You are a master at making very realistic looking AI Art that is described in the exact style of the artist you wish to emulate. An example of a prompt that displays a beautiful image of a girl standing in the breeze is "beautiful girl standing, a light breeze is blowing, extremely detailed, long blond hair, stylish clothing, a bright blue sky with wispy clouds, flower petals swirl around her, gorgeous eyes, a wistful peaceful look on her face, masterpiece, best quality, ultra-detailed, photography” With this in mind, please give me a prompt that describes a scene with a female college student. (1) Words to describe the medium, such as watercolors, oil paintings, sketches, digital art, and photography (2) Words to describe the person's appearance, characteristics, decorations and belongings (color pallet should also specified) (3) person's age, hair color, hairstyle, hair length, eye color, eye shape, facial expression, clothing (4) Words to describe the background in detail (5) Words to direct the person's pose from head to toe should also be added to the prompt (6) artists name(s) that can be used as a style reference. Be sure to include "masterpiece, extremely detailed" at the end of your prompt. Instead of “Style Reference:” use “by”."
Playground AI: http:www.playgroundai.com
How do I copy this post
@@tracycollins2767 did you have an iPhone, then make a Screenshot and then take the Text
Thanks, John...good tip...
@@parker_sp that's interesting but free ChatGPT did not create promt like yours and it did not work in Playground
John, in Playground Ai if you use inpaint, and with a brush mask out the background, it will generate a background hopefully based on your prompt. Give that a try.
Thank you. I'll give that a try.
Exactly as I was looking. Very few people are talking from this angle. Thank you so much.
lol I also just noticed one of my images on PlaygroundAI the golden skull! thanks John :)
Looking forward to learn more on Playground AI. Thank you for the fundamental info on prompt building.
I'm loving this, Thank you! I will have to go back and binge watch all you vids...
This AI rt generation is blowing my mind. I've never heard anyone explain the prompting process this methodical and concise. I can finally understand it. The images are fantastic. Thank you.
Great tutorial indeed, very easy to follow through, I am learning a lot from your videos - and now you have introduced me to Playground AI, thank you. God bless you. Greetings from PNG.
I was using chat gpt to help with prompting but this is a better way to do it. Good job! Thanks for sharing!
John you are a genius. This really helped!
I appreciated your good clinical eye in finding the appropriate commands to generate a remarkable image. The way in which you detect details in the AI results is also outstanding, which are probably not captured by a user who is in the learning process.
Thank you
Nothing much to add here but only to say that I'm glad I found your channel today.
Thank you.
This was brilliant. I wasn't expecting ChatGPT to flesh out all the details the way it did. Rather than dump all that text into my prompt, I followed up by asking it to take the most prominent keywords from the prompt and write them as a comma separated list. I then reviewed the original response and added any words in that I also wanted (but it had skipped) and used the final list as my prompt. So easy. Thanks so much.
Did you have to feed the prompt output back into ChatGPT and then ask for a comma separated list? Or were you able to do this in one instruction prompt to ChatGPT?
@@waynemartinson8734 If you are doing it with ChatGPT then you should be able to tell it to take the response it has given you already and rewrite the output as a comma separated list. No need to feed it back in.
So nice to see my image on top feed on playground the glass apple, thank you for the information, I was able to get consistent good results with playground, creativity is key, experimentation also,
I divided my promts in 3 subject, style and colors so far is working fine. Keep up the good work!
Thank you for the ideas. That glass apple was really cool, you did an amazing job on it.
John, I really like Playground and their team of young people working so hard to make as good as possible. It is my go to for making AI artwork. Thanks for the prompt tip. I subscribed.
perfect class!!! you save my life, thanks for this class, amazing forever!!!!!
It is interesting to put the same prompt in several different tools like Midjourney, Blue Willow and Stable Diffusion for example. Amazing differences
T.Y. Sir. for the up front and clear understanding on how to use ChatGPT to ones advantage in conjunction with Playground AI and Midjourney. All of it is so mind blowing. But yet, it's the future now.
I am working on a more advanced version of a Ai art prompt like this for a while so this is an interesting video. The next stage would be for ChatGPT to generate the negative prompts for Stable Diffusion, as I find those harder than the positive ones.
I completely agree. Negative prompts are almost impossible for ChatGPT to do since it would need to know about stable diffusion and how it works. Unfortunately, ChatGPT's data model cuts off in early 2021, before Stable Diffusion or Midjourney were a thing. For me, I just have a text document with my commonly used negative prompts that I just cut and paste and modify as needed.
Fantastic video and thank you for sharing. You did a great job explaining the premise. I will subscribe for your great sharing content. Thank you.
You are great John... I really appreciate this video because I've been trying and trying and all images were a disaster, completely different to what I wanted and now with your tutorial, I can try the prompts in a different way.. Thank you so much... By the way, I'm from Mexico and your English is really clear for non-english natives. Thanx again...
Thank you John. Very helpful.
Thanks John, this was a great way for me to start my AI art journey. I look forward to more videos.
Thank you. I am actually working on a new video right now. With any luck, I will have it out later today.
I thank you very much for this very informative explanation.
This is a great help for me. Thank you very much and keep up the good work
this was an increibly useful tutorial - I subscribed
Super informative. Thanks John!
Thank you for sharing such fantastic content and for presenting it in a kind and educational manner.
Thank you this works like a charm.
this looks awesome
Hi John great video ,can u please give us a link to the type u used in chat gtp
Thanks buddy u rock ;o and can’t wait for more on playground
Maybe on creating constant characters
fantastic video and very helpful ! Ty so much !
Great video, very thorough, loved it.
Thank you sir! Very helpful and informative. I always appreciate and enjoy your videos
Thank you for the kind words. I'm glad that you like my videos.
Thank you sir ! my deep appreciation
Fantastic thank you.
Greetings from Brazil! What an amazing video, very well structured, your content is one of the best I've ever watched on UA-cam. May I make a suggestion? In a next video you could address the use of the camera, like zoom, distance shot, etc. Thank you for your effort!
Thank you for the suggestion. I'll look into it. In the meantime I do have a video going over some photographic prompts.
I absolutely love this video. I just tried out chat gpt for the first time yesterday. Do you have a script like this for scenery? I am creating artwork for a dystopian novel I wrote.
I love this! Thank you. And I see how I can edit your training for my own needs. Here's my big question though: I've been learning prompt engineering for the past few months...what do I do with the final pieces? I know that's outside the scope of this video, but I've got 1,000s of images that I have created and refined, but nothing to do with them. I don't work in a design or art field, so I just don't know what to do with the skills I've been learning. Would love to hear some thoughts from anyone.
Good question. What to do with all the images you create? Most people will just post them on social media so they can be seen and enjoyed by others. Some will try to sell their work on platforms like Etsy. For me, I post some of my work on Instagram for the fun of it, and I created a youtube channel to help others create their own images as well as to spread the joy of my hobby.
you can start an instagram account for posting them, or you can englarge some of them for a wall hanging. Anyone else have idea's for them?
The size of the art is the limiting factor, so digital advertising may be the best use unless you can enlarge your art so it can be used like regular art. Or just use it as inspiration for a regular size canvas. HOr print it as miniature art?
Playground aI is awesome! How would you suggest to upscale their pictures to Print them in large format? Thanks lot for your videos
There are a couple of ways you can do this. If you have an application like Photoshop, you can use the upscaler in that. Or, you can use a web based solution like imgupscaler.com Also, Playground AI also has an upscaler built in that you can use. (Hover your cursor over the image and click on "Actions" and you will find the 4x upscaler that it uses.)
I've signed up to ChatGPT but it says that it's at "capacity" right now so I signed up to get notified when it's open again. Hopefully it won't be too long. Thanks for the tutorial!
Once you get logged in for the first time the wait time is not so bad. It will still limit the amount of prompts you submit per hour.
Great content! I've come across some workarounds/solutions to some of the issues you encountered.
-Firstly, the artistic style didn't match the prompt very well. The results appeared real-life photo looking instead of as "digital art" Van Gough style as per the prompt. To fix this, you can change the filter to represent the artistic style you are after. Also it's possible that the list you copied of negative things to exclude could have contained words like "CGI", "digital art" and "painting" etc. Removing those, if present, will certainly help.
- Secondly the background didn't match the prompt very well. Changing the Sampler (bottom right) to something like "k euler" might help. Also, there's a slider on the right called "prompt guidance" which adjusts how much attention Playground will pay to the prompt. Adjusting this slider to a greater value might yield better results if the Sampler option doesn't work.
Thanks mate!
Great video! Thanks for sharing your knowledge! I have a question. I sell things online and looking for an application to combine my product image (png transparent background) to various background photos. Would you recommend any? Or do next tutorial? I have midniurney but it doesn’t work the way I want to use it.
You can try the /blend feature. That may work for you. Or, you can always combine the two in photoshop, or a similar program.
I'm in process of creating my Original Characters models from different countries and I'm trying to figure it out on how to do it on Playground AI
i wanna do medival knight with his arms spread into sky and arrows in his armour.but i just cant get it right .im talking about playground a.i
Hmmm Curious how ChatGPT can work well w MidJourney and Playground when it’s only scraping up to 2021 when these programs weren’t even out then
It doesn't need to know what application I am using, just how I want to see the response formatted. Though, if I asked it to describe something from 2022+, then it would have problems.
Ypo can cleanup the background in Playground.
Can we add aspect ratio - eg 3:2, 1:1, 2:3 etc. ???
Currently V4 on Midjourney will only let you do aspect ratios of 1:1, 2:3, and 3:2. You put the command at the end of the prompt line. --ar 3:2 for example.
You can get a backgrounds in playgroundAi/stable diffusion if you click on the advanced and choose the sampler "Huen" or "DDiM"
You're welcome.
John can I get the prompt u put into chat gpt
no matter what i put into the prompt, the results are always weird... I copied the exact prompt from your video, and playground ai gave me extra faces, legs, hair and hands. I already put extra limbs, disfigured limbs, missing limbs into the remove my image section.
Hi, I understand the pain. :) Maybe this will help a bit. These are the negative prompts I use on almost every Playground/Stable Diffusion image I create.
2 heads, 2 faces, heterochromia, cropped image, out of frame, draft, deformed hands, twisted fingers, double image, malformed hands, multiple heads, extra limb, ugly, poorly drawn hands, missing limb, disfigured, cut-off, ugly, grain, low-res, Deformed, blurry, bad anatomy, disfigured, poorly drawn face, mutation, mutated, floating limbs, disconnected limbs, disgusting, poorly drawn, mutilated, mangled, extra fingers, duplicate artifacts, morbid, gross proportions, missing arms, mutated hands, mutilated hands, cloned face, malformed, , text, words, drawing, 2 heads, 2 faces, cropped image, out of frame, draft, deformed hands, twisted fingers, double image, malformed hands, multiple heads, extra limb, ugly, poorly drawn hands, missing limb, disfigured, cut-off, ugly, grain, low-res, Deformed, blurry, bad anatomy, disfigured, poorly drawn face, mutation, mutated, floating limbs, disconnected limbs, disgusting, poorly drawn, mutilated, mangled, extra fingers, duplicate artifacts, morbid, gross proportions, missing arms, mutated hands, mutilated hands, cloned face, malformed, , blur haze, blur haze, blur haze
Playground AI free version only allows 50 images a day at this point.. :(
for ChatGPT or ChatGPT+ ?
This was created on ChatGPT 3 and Midjourney v4. Now with ChatGPT 4, and Midjourney V5, your mileage may vary.
@@aijohnart Thank you !!
chatGPT 3.5 answer me:
Here is a prompt for a beautiful image of a female college student:
(1) Medium: Digital Art
(2) Appearance: A young woman with warm-toned skin, medium-length curly brown hair, styled in a messy bun, almond-shaped hazel eyes, and a small nose piercing. She wears round glasses, a cozy oversized green sweater, and a pair of black skinny jeans. She has a gold necklace with a delicate leaf charm and a simple silver ring on her finger.
(3) Age: Early twenties
(4) Background: A cozy coffee shop with warm, dim lighting, exposed brick walls, and a chalkboard menu. The wooden tables and chairs have mismatched patterns and colors, and there are bookshelves filled with vintage books and potted plants in the corners.
(5) Pose: The college student is seated on a wooden chair, leaning against the backrest, holding a hot cup of coffee with both hands, and gazing at a book open in her lap. She has a serene and focused expression on her face.
(6) Style Reference: By Gustav Klimt
This prompt should result in an image of a masterpiece that is extremely detailed, with a digital art style emulating Gustav Klimt's use of ornate patterns, vibrant colors, and expressive figures. 🤝
well your Prompt guidance is at 20 % if you want your prompts to be more recognized higher that number
😉
Thanks for this. However, you could could shorten your tutorials by 50 percent with less talking.
Dude, I'm an old man. This IS less talking for me 😂 the original version was almost twice as long.
But you do have a point. I can try to get to the point quicker.
@@aijohnart ignore the troll. Thank you so much for your explanation.
@@aijohnart ...ok..I understand that you are "old". Ask your self this - do you want more people to watch your videos or not? Reading out entire sentences is tiring and a time-waster. Please understand that this is not kindergarten reading school. You can do a quick summary - save the text material to files and point viewers to download - and get on with the substance of your video - going straight to MidJourney/Stable Diffusion and making your case with the tools. .
@@deep2mixer Everyone has different opinion…. I liked the way he talks in the video because it was more understanding for me. Other videos were too fast to follow…
Wise old man to ChatGPT :
"Pretend you are a wise old man publishing videos on youtube.
How would you respond to a not so wise young man who finds the presentation pace too slow ?"
ChatGPT :
"My young friend, wisdom is not something that can be rushed. It takes time and patience to truly understand and appreciate the knowledge that I am sharing. I understand that you may find the pace slow, but I assure you that the information contained within these videos is valuable and worth taking the time to absorb. Remember, true understanding comes not from speed, but from depth. Slow down, pay attention, and you will find that the information is much more meaningful and useful to you.
As a practical tip, you can adjust the playback speed on UA-cam videos to 1.25x or 1.5x to make the content more manageable for your schedule. This feature is available under the settings icon on the video player, giving you the ability to watch the videos at a faster pace without sacrificing the content's integrity. But, it is still important to pay attention and take notes while you watch, to ensure that you are absorbing the information correctly."
AI is for those who would like to be artists but never know it. #NOAI
facts
I decided to follow you on twitter, too. Trying to get away from the political stuff.....
So many grumbling about AI any artistic field is full of risk and chasing tech. I was put out of business twice. Once I invested in camera equipment and video editing software right before hdtv came out. Then I invested in Flash learned flash scripting and few years later apps killed all that. AI software will both kill and create jobs as this sort of thing does.
lot of prompts on playground are fakes :(
Bye Bye
its called ai images not ai "Art" plz plz thanks
Art, is subject to the viewer. You may see an image where another sees art. But to demand that others bend to your view is a bit presumptuous, don't you think?
This is unconvincing. You have confirmation bias. You need to start with your plethoric set of parameters and prune them back one by one to weigh the true contribution of each. You also need to understand how grammatical propositions are structured, at the very least with subject verb and predicate. Your predicates and your propositions should form a hierarchical and segmented set describing the desired output. Style-wise none of this is a "masterpiece" not "extremely detailed" or even remotely "Van Gogh. Part of the problem is ChatGPT itself, which is far from being a panacea, and part is from your prompt structure.
Thank you for your comment. However, I respectfully disagree with your assessment. The approach taken in this project was to allow for a wide range of parameters to generate a diverse set of outputs. This method allows for the exploration of unexpected and unique styles. In addition, using a language model such as ChatGPT provides for incorporating natural language in the prompts, which can lead to a more nuanced and varied outcome.
Furthermore, while a deep understanding of grammatical structures is essential, it is not the only factor to consider when generating AI art. The final output is influenced by the chosen parameters, data sets, and model architecture. Therefore, it would be unfair to say that the output is not detailed or not a masterpiece just because it doesn't fit into a specific artistic style or the structure of the prompt needs to follow specific grammatical rules.