Full frontal, landscape, macro, street, parallel, horizontal, side view, back view, slow shutter speed, fast shutter speed, above and below i.e water, cave, earth, backlighting. spotlighting, sidelight, soft lighting, 8k, octane rendering, your favourite photographer's name, camera type, lens type lens refraction, miniature, utilising these and what surge has said the world is your oyster. You will be hooked just like I have since September 2022. Enjoy
I dont think MJ4 works with just typing a prompt with a scene. Ive done over 10,000 images. Have a collection of 300+ prompts. Im alwats trying to work out how to mould and sculpture. Its extremely difficult to get x3 ideas in a scene. Example, -New York street, -huge twisting roots shoot out of the ground & start to wrap around the skyscraper building. -crowd panicking run in fear. I found i could create roots shooting up and being close to buildings (very poorly & very hit n miss). Roots usually grow down, AI was getting lost and confused. But there was no way i could get the crowd , it was new york street & crowd or new york street & giant roots... but 3 of them together produced non photo realistic images. Maybe i need to do more research and find horror creators who the AI can be inspired by??
The only way I could do different elements like this was separating them in different images to relate to. In the same prompt, MJ deals okay with 2 elements + 2 IMG references, try this. [img] [img] your prompt.
@@AIArtistTV yes. X2 seems to work. I did det results using variation, but this starts to look a bit strange, the realness starts to break. Maybe that could be a challenge for you. Trying to get x3 elements with seed, variation, img prompts. i was getting results but they were very hit and miss. Another great feature i noticed was, if i wanted a huge crowd i would type in subjects where huge crowds would be seen, festival,,marathon, zombie horde, infantry, 100,000 troops. So trying to work out if i need sinister roots growing out of ground should i look up and horror films or artists who do this kind of genre. Does adding a comic book artist break the "photorealism" aspect?
@@armondtanz my main issue is not the subject itself but aspecs i.e fingers feet animals missing a leg or out of place. I read on a furom +6 finger work but not tried. Seems most are having these issues. I all I absolutely enjoy using Midjourney with my imaginality not just photo but modern art. @Paulie on midjourney.
I learned by observation of what others were doing. copy and pasting onto note the prompts that seem to work well. I know that if you put in subject background adding odac last works very well also
Here are a few fun ones to try out: "landscape, high-angle, digital, wide-angle lens, Ansel Adams, mountain, climbing, small aperture, dramatic" "portrait, low-angle, film, 50mm lens, Henri Cartier-Bresson, person, playing guitar, fast shutter speed, candid" "still life, straight-on, digital, macro lens, high ISO, glass of wine, bottle of wine, low depth of field, minimalist"
Great video and information. I have a TH-cam channel that I would like to make a custom background for and this seemed like the perfect program to do that. What would be the best prompts to have an office background or a home office background with a bookshelf or just like about around the way you have in your video but do it mid Journey thank you
Very new person here to AI, once you generate something you like with a “person” in the image. Is there a way or prompts as they called to keep that same “person” in the generated image but change different aspects of it in terms of back round and what they doing in image clothes on said “person” etc??? I’ve played around with a few things but I can’t seem to keep the image of the generated person that I want but doing what I’ve stated above. EXAMPLE, I’ve been wanting to create images of a golfer on a golf course, keep the “person”but have the location of the person change from the tee box, fairway, bunkers, on the green etc. Any help greatly appreciated. Cheers.
Very good, but I wish you start to use teleprompters or something that stays focus don't he subject without so much speech imperfections and more pause , but good video , thanks
30:49 Oddly enough, I've seen some people use 32K think that MJ would render the image is super-duper-duper high resolution. 🙄 But the results they get are consistently not higher res than if they just used 4K.
Gifts from the auto-stylize and auto-chaos when we wont specify those, same miracles from people who write a super basic prompt, like, "cat in the beach" and get an amazing result, because MJ is background is making a 5 lines prompt.. :P
@AI Artist TV Speaking of stylize, I typically add the "--stylize" switch and use a large parameter usually above 700 on most of my renders without often comparing it to no added stylize switch. I do that because at one point a while back I was getting relatively consistent good results with certain images using that "--stylize" switch. But I'm wondering from your experience if you've found using the "--stylize" switch to be helpful in any of your renders? Ideally I should just always try to compare results with and without -- switches but time is not on my side and I burn through my 15 hours of Fast time so quickly. Also, the only time I've got consistently good results with the chaos switch is when attempting to reduce the weight of the image part of a prompt. That alone may sound strange that I'd even want to reduce the weight on an image but when that image is a wild abstract twist of colors and I'm trying to get an image of a strange magical forest which has *SOME* limited characteristics of twisting colors, that's the only way I know how to in ver. 4 since ver. 4 doesn't permit weighting of image prompts.
In your prompting, you use "lenses" instead of what would seem to be a much more appropriate term, "lens". Is there a specific reason you use the plural of lens? 🤔
Man, this is curious for me too, but on my tests, it seems works better when I use lenses, I can be wrong, anyway, I am using more SLRs lately.. so I am not using much the lens...
@@AIArtistTV words are vector points in latent space, do not take it so litteral you are pointing to a area in that space composed of millions...billions of images wich correlates to that point in latent space, so it very well can be that "lenses" prodcues better result then "lens".. also depeding on the other words in your prompt, removing even adding one word will change your reesults.
Agree, but it was happening also because MidJourney was trying to understand more about the main subject, if I have specified more than just walking, they would also desapear. But you're totally right, that would be perfect if I need a man only walking. Thanks man.
Man, thank you so much! This is indeed my biggest challenge here, I have to slow down, but I think I always gett excited in my videos, but thanks man! Working on it! Don't give up on me! hehe
very interesting tips . . . thank you . . . but pls slow down as you stutter a lot and it could get a little painful to listen to your commentary unfortunately . . .
@@AIArtistTV i talk like you. I dont stutter (and you dont) but sometime re repeat certain words. Ive been surrounded by toxic ppl all my life who mock & make fun of it , so you just carry on.
[type][angles][medium][equipment][photographer],[subject][activity],[technical][quality]
for the lazy ones like myself. Great tutorial. Thank you.
Full frontal, landscape, macro, street, parallel, horizontal, side view, back view, slow shutter speed, fast shutter speed, above and below i.e water, cave, earth, backlighting. spotlighting, sidelight, soft lighting, 8k, octane rendering, your favourite photographer's name, camera type, lens type lens refraction, miniature, utilising these and what surge has said the world is your oyster. You will be hooked just like I have since September 2022. Enjoy
I dont think MJ4 works with just typing a prompt with a scene.
Ive done over 10,000 images. Have a collection of 300+ prompts.
Im alwats trying to work out how to mould and sculpture. Its extremely difficult to get x3 ideas in a scene.
Example,
-New York street,
-huge twisting roots shoot out of the ground & start to wrap around the skyscraper building.
-crowd panicking run in fear.
I found i could create roots shooting up and being close to buildings (very poorly & very hit n miss).
Roots usually grow down, AI was getting lost and confused.
But there was no way i could get the crowd , it was new york street & crowd or new york street & giant roots... but 3 of them together produced non photo realistic images.
Maybe i need to do more research and find horror creators who the AI can be inspired by??
The only way I could do different elements like this was separating them in different images to relate to. In the same prompt, MJ deals okay with 2 elements + 2 IMG references, try this. [img] [img] your prompt.
@@AIArtistTV yes. X2 seems to work. I did det results using variation, but this starts to look a bit strange, the realness starts to break.
Maybe that could be a challenge for you. Trying to get x3 elements with seed, variation, img prompts.
i was getting results but they were very hit and miss.
Another great feature i noticed was, if i wanted a huge crowd i would type in subjects where huge crowds would be seen, festival,,marathon, zombie horde, infantry, 100,000 troops.
So trying to work out if i need sinister roots growing out of ground should i look up and horror films or artists who do this kind of genre.
Does adding a comic book artist break the "photorealism" aspect?
@@armondtanz my main issue is not the subject itself but aspecs i.e fingers feet animals missing a leg or out of place. I read on a furom +6 finger work but not tried. Seems most are having these issues. I all I absolutely enjoy using Midjourney with my imaginality not just photo but modern art. @Paulie on midjourney.
I learned by observation of what others were doing. copy and pasting onto note the prompts that seem to work well. I know that if you put in subject background adding odac last works very well also
Really nice! Thanks for share! superpowers :)
Thank you! Cheers!
Thanks very much. Eye opener
Very welcome
Very helpful, thanks very much!
Glad it was helpful!
great video!
Thanks!
Here are a few fun ones to try out:
"landscape, high-angle, digital, wide-angle lens, Ansel Adams, mountain, climbing, small aperture, dramatic"
"portrait, low-angle, film, 50mm lens, Henri Cartier-Bresson, person, playing guitar, fast shutter speed, candid"
"still life, straight-on, digital, macro lens, high ISO, glass of wine, bottle of wine, low depth of field, minimalist"
Wow! I like all of them :)
Thank you - your content is exactly what I am looking to learn!
Great to hear!
thanks its very interesting technical
There you go! :)
Hi, Please share Angles and scenes excel sheet. you taught in camera, angle and distance subject video
Great video and information. I have a TH-cam channel that I would like to make a custom background for and this seemed like the perfect program to do that. What would be the best prompts to have an office background or a home office background with a bookshelf or just like about around the way you have in your video but do it mid Journey thank you
you can use "low light HDR" in the prompt to counteract the punchy contrast. I have no dea if they just cancel each other out, but it does work.
wow! good to know! will test asap!
Fantastic tutorial, thanks so much for taking the time to do these.
You're very welcome!
Very new person here to AI, once you generate something you like with a “person” in the image. Is there a way or prompts as they called to keep that same “person” in the generated image but change different aspects of it in terms of back round and what they doing in image clothes on said “person” etc???
I’ve played around with a few things but I can’t seem to keep the image of the generated person that I want but doing what I’ve stated above.
EXAMPLE, I’ve been wanting to create images of a golfer on a golf course, keep the “person”but have the location of the person change from the tee box, fairway, bunkers, on the green etc.
Any help greatly appreciated.
Cheers.
25:08
😆 🤣 😂
That's one giant lens!!!
😂
Thank you so much for this. So much information! I can't wait to see if it recognizes Lensbaby and Kodachrome 64.
Haha! Just did on video #009 :)
@@AIArtistTV I'll check it out. Thanks!
how i can sell my photos sir? any idea or trusted sites ?
Hi my friend! I like a lot to use wirestock.io/ since it will distribute for you on many websites. :)
Why don't you use a seed number to make the image consistent?
Very good, but I wish you start to use teleprompters or something that stays focus don't he subject without so much speech imperfections and more pause , but good video , thanks
30:49
Oddly enough, I've seen some people use 32K think that MJ would render the image is super-duper-duper high resolution. 🙄 But the results they get are consistently not higher res than if they just used 4K.
For some people, prompt is a "leap of faith".. :P
@@AIArtistTV
😆 well put!!
28:00
That's interesting that you didn't even prompt for "bokeh effect" and yet MJ automatically rendered that effect (or semi bokeh effect).
Gifts from the auto-stylize and auto-chaos when we wont specify those, same miracles from people who write a super basic prompt, like, "cat in the beach" and get an amazing result, because MJ is background is making a 5 lines prompt.. :P
@AI Artist TV
Speaking of stylize, I typically add the "--stylize" switch and use a large parameter usually above 700 on most of my renders without often comparing it to no added stylize switch.
I do that because at one point a while back I was getting relatively consistent good results with certain images using that "--stylize" switch.
But I'm wondering from your experience if you've found using the "--stylize" switch to be helpful in any of your renders?
Ideally I should just always try to compare results with and without -- switches but time is not on my side and I burn through my 15 hours of Fast time so quickly.
Also, the only time I've got consistently good results with the chaos switch is when attempting to reduce the weight of the image part of a prompt. That alone may sound strange that I'd even want to reduce the weight on an image but when that image is a wild abstract twist of colors and I'm trying to get an image of a strange magical forest which has *SOME* limited characteristics of twisting colors, that's the only way I know how to in ver. 4 since ver. 4 doesn't permit weighting of image prompts.
In your prompting, you use "lenses" instead of what would seem to be a much more appropriate term, "lens".
Is there a specific reason you use the plural of lens? 🤔
Man, this is curious for me too, but on my tests, it seems works better when I use lenses, I can be wrong, anyway, I am using more SLRs lately.. so I am not using much the lens...
@@AIArtistTV words are vector points in latent space, do not take it so litteral you are pointing to a area in that space composed of millions...billions of images wich correlates to that point in latent space, so it very well can be that "lenses" prodcues better result then "lens".. also depeding on the other words in your prompt, removing even adding one word will change your reesults.
One way to rid cameras in the subject use --no camera tag
Agree, but it was happening also because MidJourney was trying to understand more about the main subject, if I have specified more than just walking, they would also desapear. But you're totally right, that would be perfect if I need a man only walking. Thanks man.
Bro, valuable information. Thank u
Buuut: speak slowly, speak less,
Speak clearly! Ur English is hard to understand.
Or use Use Murf AI!
Man, thank you so much! This is indeed my biggest challenge here, I have to slow down, but I think I always gett excited in my videos, but thanks man! Working on it! Don't give up on me! hehe
very interesting tips . . . thank you . . . but pls slow down as you stutter a lot and it could get a little painful to listen to your commentary unfortunately . . .
Thanks for the tip! I agree I was kind too fast on this one... Thanks for pointing out, will start to be more careful about it. :)
@@AIArtistTV no worries - Thanks for understanding . . . I hope I did not come out too harsh on you :) All the BEST !!
@@ApolloReyes OMG! Of course not my friend, you stopped by to suggest something to make me better! hehe.. I thank you. :)
@@AIArtistTV i talk like you. I dont stutter (and you dont) but sometime re repeat certain words.
Ive been surrounded by toxic ppl all my life who mock & make fun of it , so you just carry on.
Ai cant make a realistic photo no matter how you try you will always know its not real, but in the future maybe.