This Chat GPT hack will change your midjourney game | How to create realistic interior images
Vložit
- čas přidán 25. 03. 2023
- I came across this thread from @nickfloats and it helped me get amazing results out of midjourney using this awesome chat gpt hack. I was really struggling to get great images out of midjourney but after I read this thread, the images that I have been generating have been crazy good! Also, midjourney v5 that was recently released has been a game changer too!
Here is the original thread from @nickfloats :
/ 1635116676978208769
Image prompts:
Image 1:
Editorial Style photo, Low-angle, Bohemian, Living Room, Textured Rug, Plants, Pillows, Decorative Objects, Warm, Earthy Tones, Anthropologie, Soft, Diffused, Marrakesh, Morocco, Afternoon, Bohemian --ar 16:9 --stylize 1000 --v 5
Image 2:
Editorial Style photo, High-angle, Industrial, Loft, Exposed Brick Wall, Metal, Concrete, Furniture, Cool, Neutral Tones, West Elm, Soft, Directional Brooklyn, NY, Morning, Urban, Industrial Loft --ar 16:9 --stylize 1000 --v 5
Image 3:
Editorial Style photo, Eye-level, Mid-Century Modern, Home Office, Desk, Chair, Velvet, Wood, Metal, Desk Accessories, Bold, Primary Colors, Herman Miller, Soft, Directional, Los Angeles, CA, Morning, Productive, California, Modernism, --ar 16:9 --stylize 1000 --v 5
Image 4:
Editorial Style photo, Low-angle, Minimalist, Bedroom, Bed. Concrete, Linen Bedding, Accessories, Black and White, Muji, Soft Lighting, Tokyo, Japan, Evening Calm, Japanese Minimalism --ar 16:9 --stylize 1000 --v 5 - Jak na to + styl
Please Create a table that breaks down an interior architecture photograph composition into the follwing key elements , where each of these key elements is a column : composition , camera angle , style , room type , focal point , textures , detail , color palette , brand , lighting , location , time of day , mood , architecture
Fill the table with 10 rows of data ,where :
Composition = "Editorial Style photo"
God sent Angel! :)
God sent
Thank you
This blew my mind. I’m so so excited.
Thank you so much for sharing this incredibly, valuable and helpful content!!! I can't wait for more videos.
So glad you found this helpful!
Thank you so much, it helped me a lot with my midjourney interior design mockup generation.
Glad it helped!
omg this is VERY helpful. Thank you so much for sharing. I use that with DallE and it levels up the quality of the image SO MUCH!
Thank you to you too for expanding on Nick's idea ❤
😘😘😘
Great video, subscribed, super concise and direct!
Amazing, thank you! 😊
Really informative, thank you for sharing!
You're welcome!
awesome vid and nick floats is great!
Thank you and he is!
This channel is gold! The content your videos are so well organised, effective and short. I love your exploration of how designers can make use of various AI technology, its a path I'm keen on exploring as well. Feels like I'm going on a journey with you :D Looking forward to your next videos!
Aaaaahhh thank you!! That is so wonderful to hear!! After putting in so much effort into these videos it's nice to hear comments like this 😊
Really like that❤
brilliant!
Thank you, I'm just the messenger though 😁
Amazing channel!
Hello from Romania! You look amazing and I follow you in everything you do, I would like to have as much information as possible about AI, how you insert your own image and how you modify it, what exactly is your workflow! Anyway, thank you for existing and making our lives easier! :)
Thanks for following along and the lovely compliments! I'm not really that well versed in AI but learning along the way. I've got a video coming out soon that touches on mj settings and prompts and how you can get better outcomes. Also, I'll briefly touch on how you can insert an image and make variations on it too 😊
@@theredonechannel I will wait with the greatest interest.
Fill the table with 14 rows of data ,where :
Mobile phones names on each row of these phones : moto edge 50 pro, oneplus 11r ,Samsung A54 5g, pixel 7a, samsung F54 5g, Vivo V29, samsung S21FE, vivo v30 ,samsung s20FE , Pixel 6a, nothing 2, note 13, realme12pro ,iqoo neo9 pro .
Hi this is really awesome! Thank you! Is there a way to use exact measures form a floorplan or 3d model to have controll over how the room will look? (so the measures arent randomly generated).
Also if I want the same room from a different angle?
Thank you!
Unfortunately midjourney isn't very good at that. But you can use a program called coohom which is a very easy 3d program to use! And it's free up to a certain point!
Hey there, is there a way for MJ to produce the correct sized frames when I ask it to? For example, I'll ask for the dimensions to be 4x5, and it'll totally ignore me and do maybe like 13x17 or something.
Thank you! I think it's light years ahead of Dall-E
It definitely is!
😶🌫the images are crazy (even if some details are not perfect) !!!
I just made a test with Adobe Firefly with the same prompt : we're close to a catastrophe: the result isn't nearly as complete, and the rendering is... ugly.
Lol, midjourney is still the best at generating images by far!
Can you send IA a vision board with materials etc and receive a 3d visualization based on it? Is it possible in any of those programs?
wow! is possible to import a photo of an empty room and use this workflow ?
g8 video!
Can you keep the style that you like example in PHOTO 1 (living room) to create the bedroom? bathroom?
Yeah totally! Just change to word 'living room' to 'bedroom' or whatever room you're after, in your prompt 😊
HI! This video was so helpful, thank you. I have one question though: do you know how to get midjourney to add a specific piece of furniture though? like through a website link or product link maybe?
agreed! thanks so much for making this video. i would love to know this too!
Hello - thx so much for that information - it's really helpful. I got one question I couldn't find a useful answer till now: How can I include a kind of dimensions, such as width and height or number of storeys? When I ask for a two-storey family villa, it's more or less ok, but there are always some images with much more - the same happens when I ask for a narrow construction. Can you help me with that please?
MJ isn't very good at following dimensions, it's not quite at that level yet. You can try using different terminology like double story instead of two storey? Sometimes playing around with similar words can change the results 😊
Do you know how I can input my interior layout plan into AI software and create 3D renderings according to my prompt? such as modern style, French Provincial, etc?
Unfortunately, I don't think midjourney is capable of this just yet. There might be alternative AI software out there that may be able to do this but none that I am aware of!
i accidentally came across your video and its amazing!
can i ask you, i have an eshop of men and women hats and my main pain point is the lifestyle photos i have to take to present the hats.
DO you think i could somehow upload the photo of a hat and showcase it on a table, a chair, a desk or even to a humans head?
If so, could you guide me on how to do it as i am new to all this chat gpt and midjourney?
Thanks!
Panos
Unfortunately midjourney is not that advanced yet. You can upload an image into midjourney, e.g like your hat and then ask midjourney to place that hat into lifestyle shots but it doesn't quite work that well (i did a couple of tests). It can't take that hat and place it perfectly like on a person's head or on furniture. It tends to distort or reimagine the hat for some reason!
@@theredonechannel maybe then i could place it on e.g. a table. how would that work?
Could you kindly share the 2 phrases for me to tryout? i would be grateful!
@@theredonechannel btw, did you try on Midjourney with V5?
@@Giantpanos I tried to upload an image into midjourney and asked midjourney to place that hat on a table but it didn't work out well. It deformed the hat and did some strange things but couldn't really follow instructions that well. So unfortunately I don't have any good prompts for you that would be helpful!
@@Giantpanos Yes, I did use V5 😊
Is there a way provide MJ with a photo of your current space and ask it to reimagine it using the information in your prompt? That kind of feature would be a game changer for interior design, especially if all of the items added to the final image were linked to products currently available for purchase.
I've tried a couple of tests but it's not able to put everything into the perfect place, with the exact piece of furniture...yet. I reckon we will see huge improvements in the program in the next couple of years. There are a couple of basic programs you could use to change items of furniture around in your space but it's very basic. The one is called reimaginehome.ai
Did you try training with your image URL? I had to go back and forth between GPT and MJ multiple times before my image started to resemble the URL image. It's still not perfect, but its getting close.
@@JaySpoonn I haven't actually! When you say training, do you mean like practicing?
@@theredonechannel You could say it's a form of practice, but when I say train I mean Chat GPT learns and remembers the content of the conversation in each individual prompt that you set up. I.E you make a description for a room, and want to use GPT to expound on this description. You plug that description into MJ for instance, and it doesn't look exactly how you like. The more you do the process of using GPT to describe your prompt the more refined the descriptions will become. Until you reach the plateau of prompts for said description.
@@JaySpoonn Oh yes, I have heard about this but have never actually tried it! Will definitely remember this next time, thank you 😊
Is there a way to upload an image into MJ and have the AI generate a scene around it? For example - if I have a specific product/style/framework I want to build around, can I upload that image into MJ and have the AI generate a scene around it? I'm new to the channel & I love the content! Very helpful!
Hi Tyler! You can upload an image into midjourney but the zoom out function (which is the function that generates the scene around) only seems to be available when you generate an image within midjourney. So if you upload an image and then use a prompt to alter the image, that would give you the zoom out function. But the chances of midjourney changing that image from your prompt is quite high so your regenerated image might look quite different from the original uploaded one. I hope that makes sense! Dalle also has that function where you can upload an image and then regenerate a scene around it, and I think you can do it from an uploaded image without changing it. I might be wrong but give it a go! I hope that was helpful and glad you find my content helpful 😊
how can I tell midjourney to leave the photo I upload as is just add those furniture? is it possible ?
Unfortunately MJ isn't that advanced yet. It can't follow instructions that well. I've tried similar prompts, like changing certain furniture elements but it isn't capable of following instructions well!
Great video, I noticed you used side-angle-views in a prompt but this didn't seem to work fopr me. All my mockups are from a head on front view and I've been looking for a way to get 45' or side views for a while now with no luck, would anyone happen to know how to get angled views in MJ ?
Sorry for the late reply! I almost always use editorial style photo and that gives me that side angle view 😊
@@theredonechannel Thank you so much, I'll give that a go. Much appreciated.
This seems to be a South Africa accent. Hi from Cape Town
It is indeed! Good pick up 😁😁😁 Hi back 🤗
Thanks for this! For commercial / business purposes with private use (images created not known to public), should I get the $30 or $60 plan? Does creating images with $30 plan in a private discord server with MJ bot helps in making all images non public and totally private? Need some advise. Thanks!
Tbh, I'm not well versed on that side of MJ so I can't give valuable feedback on that unfortunately!
Only the $60 plan gives you privacy and prevents your generations from being shown to others members. Creating prompts in your own server only keeps you from having to create in a crowded thread where your prompts and results get quickly buried.
is version 5.1 even better??
Sorry for the late reply! I believe 5.1 gives less of that MJ aesthetic and more of a realistic look 😊
Listen!!!! RSA in the house...
Lol you know it!!
Immigrated to the States when I was 10. At the core I'm South African. Recognized your accent right away.
@@bogiev1 the accent never goes away when you emigrate as an adult!
I like that you are open to AI. Also, interesting combination of ethnicity and accent.
I feel like you gotta be open to AI so you know how to pivot for the future. I am a chinese South African living in Australia so that may explain the accent lol
poor 3D visualizers....... just sad
Poor copyrighters, artists, illustrators, song writers, all jobs that could get replaced with AI. It is moving in that direction but the best way to be prepared is to know the program so you can pivot and not become redundant 😊