I’ve been using MJ for over a year (not a beginner) and still learned from this video. You should remove “beginner” from the title (and future titles) and say “tips and tricks” instead so non-beginners will click and watch. I love your videos and buy all your PDFs so I hope you will continue sharing your insights and never get bored. Much respect.
Great to hear you got something from this video, and I appreciate the support👍. I do need more testing with my titles and thumbnails, it's hard to find the right combinations that work well. Sometimes it takes a while before you know how well a video will do.
Hi Tao. Cool vids. Keep 'em comin! You asked about the Describe command with website. What you do is click the + sign next to imagine.Then hover over the image you want described, click the little "i" icon next to the bin icon. Then you wait for a few seconds (it takes a while), then you get broken down bite sized prompts.
Great tips! Thanks! I was wondering if, in 10:10, feeding the final image you generated back into Midjourney as an image prompt, maybe with an --iw 2, and making appropriate adjustments to the text portion of the prompt, would have helped refine the image. Also, in 7:48, using the original image as an image prompt, together with the prompt generated with /describe and adjusting --iw (maybe even in conjunction with --sref and --sw) would probably also give good results. Thanks again for taking the time to share your knowledge of Midjourney with us. It’s not easy to come accross chanels that are showing more advanced Midjourney techniques like yourself.
Hey, for arranging subject locations in a scene, I did try playing around with using what I generated as a reference image but it still seems really hard to get the girl placed in the top right and the boy in the bottom left, even with a high reference image weight. It usually to rearranges them onto the same elevation. Even when they are arranged correctly, it still has blurry faces unfortunately. For the Ai history ones, you could definitely also use an image reference + style reference to get it even closer than what I had in the video.
Awesome prompt tips Tao brother and appreciate you featuring my inquiry, I hadn’t thought of your suggests so def trying out BIG UPS! Also great techniques for beyond beginners indeed 💥⚡️👏🏻💯
Just watched a bunch of your videos -- so helpful. The big thing I'm working on is trying to create multiple images with the same characters in the same location but with different story beats -- aka make a comic book. Thanks for any help!
Getting the consistent scene is tricky. You could generate a background scene and then use the vary region inpainting tool to add in different characters into that same scene. It is hard to get that same scene from different viewpoints though.
Great videos. Concise. Keep going. The /describe in the new site is a little sign option on thumbnail images you uploaded., delete and describe as overlays.
Another great video! Thanks for this! For a future video could give hints on generating images of persons for character reference that doesn't being look odd? All my attempts end up with persons with eyes widely separated, or extremely oval faces.
Can you do a video on create TEXT in MJ? It's constantly creating mispelled words... and how to use the VARY (Region) to fix them - what are the proper prompting editing/adjustments, etc.
I haven't used Magnific before, I've tried Upscayl though and it works pretty well. It's also free to use which is nice, although the quality may not be as high as Magnific. I've uploaded a tutorial for that here: ua-cam.com/video/BRx9OiBOUWE/v-deo.html
Great channel, keep it up. Could you do a full movie scene in pika? It should now be possible and I'll be trying myself soon to build a short movie for my channel, want to see how our workflow would compare.
Hello Tao, how are you? Sorry if my translation isn't good. How do I generate images in Mid with 1920x1080 resolution. I tried old commands but they don't seem to work anymore.
You can add --ar 16:9 to the prompt for that aspect ratio and the use upscale on the resulting image. This should produce an image with resolution 2912 x 1632 and you can use any free image editing software to resize that into 1920 x 1080.
Hi man .I found your channel on UA-cam ,and want to say thank you for posting this helpful material .But i don’t know why I can’t get result with this method.Every time MJ creates a different person.Maybe I should use discord,exept a web version?
@@taoprompts 1-create portrait foto 2-cref, cw 100, style raw 3-mixing different types of shots and every time the shape of the eyes, lips, eyebrows and other small details changes. Perhaps MJ algorithms are not ideal yet. Maybe stable diffusion works better with characters?
@@tryzem Midjourney will be able to replicate the main characteristics but a lot of the times, the details will not be the same. Stable Diffusion + LoRA or InstantID may work better if you want the specific details down. I'm not sure if those are good with multiple character in one image though.
I'm more comfortable with discord because I've used it for so long, but I think I'll move to the website soon, it is much faster to search and organize your images.
I can't solve a problem for a long time.when I make short videos for example I create a woman picture.a girl with long hair is sitting in the forest.in runwayml in the motion brus tab I mark only the hair of the girl.so I want to move only the hair.I do ganerate but even the girl in the picture moves.i want to move only the hair of the girl.i mean i want to make it like a light breeze of wind.with which application do they do this.i create the pictures in midjourney.the ancient greek girl is sitting in the forest.in the video i want to make only the hair like a light breeze of wind and i do it in runwayml but when i create the video everything moves not only the hair.can you help me with this?
I just started using mid-journey again after about a year of discontinuing and I must say, I'm incredibly disappointed. It's amazing how little progress it has made over the past year. there seems to be definite problems that it is completely unable to solve.
Your education is very good, I wish you could tell it differently. For example, if you could make a famous actor any hero. different different scenes. this is the part I have the most difficulty with. for example arnold schwarzenegger in the role of an ancient greek hero. it would be very good. because although I tried hundreds of times, I could not succeed unfortunately. in one scene Arnold is fighting with a lion. in the other scene the same character is fighting with a dragon. if you make a training video for this, we would be very happy.
Do you mean you mean, it's not able to follow your prompts correctly, so when you ask for Arnold to fight a lion or dragon it's unable to consistently produce the same types of images?
@@taoprompts For example, I depict Arnold as an ancient Greek warrior. For example, when he goes through the forest, he meets a lion and fights it. When I type prpmpu and create the image, sometimes two Arnolds appear, one has the head of a lion, the other image has a lion and Arnold, but Arnold has a tail.
@@aresleoric7455 You could try to use the vary region to and inpaint the part of his body that was generated incorrectly. Like selecting the tail with vary region and then inpainting it with some background. Midjourney does mix things up sometimes though. Usually by reprompting and remixing you can get it to look much better.
Is there anything you would do differently?
If there was something you struggled with, someone else probably also has the same problem 🙏🙏
thats so cringe when somebody ask "may I ..." and you answer with "yes you may ..." LOOOOL
Using the camera you wanted paid off!
@@amandamate9117 I think he was just trying to be polite, it's always good to get questions in the comments.
@@amandamate9117 That's his ironic SOH. Dow!
I’ve been using MJ for over a year (not a beginner) and still learned from this video. You should remove “beginner” from the title (and future titles) and say “tips and tricks” instead so non-beginners will click and watch. I love your videos and buy all your PDFs so I hope you will continue sharing your insights and never get bored. Much respect.
Great to hear you got something from this video, and I appreciate the support👍. I do need more testing with my titles and thumbnails, it's hard to find the right combinations that work well. Sometimes it takes a while before you know how well a video will do.
Hi Tao. Cool vids. Keep 'em comin! You asked about the Describe command with website. What you do is click the + sign next to imagine.Then hover over the image you want described, click the little "i" icon next to the bin icon. Then you wait for a few seconds (it takes a while), then you get broken down bite sized prompts.
Oh wow that's kind of cool actually, it's nice that they are trying to break the prompt down a bit more.
You are incredible! And so engaging, funny, clear.
Thanks Angela! I try to make these videos as easy to understand as possible, since Midjourney is pretty complicated to use.
Brilliant - once again, you're the MJ maestro!
Thank you, I always appreciate your kind comments!
Great tips! Thanks!
I was wondering if, in 10:10, feeding the final image you generated back into Midjourney as an image prompt, maybe with an --iw 2, and making appropriate adjustments to the text portion of the prompt, would have helped refine the image.
Also, in 7:48, using the original image as an image prompt, together with the prompt generated with /describe and adjusting --iw (maybe even in conjunction with --sref and --sw) would probably also give good results.
Thanks again for taking the time to share your knowledge of Midjourney with us. It’s not easy to come accross chanels that are showing more advanced Midjourney techniques like yourself.
Hey, for arranging subject locations in a scene, I did try playing around with using what I generated as a reference image but it still seems really hard to get the girl placed in the top right and the boy in the bottom left, even with a high reference image weight. It usually to rearranges them onto the same elevation.
Even when they are arranged correctly, it still has blurry faces unfortunately.
For the Ai history ones, you could definitely also use an image reference + style reference to get it even closer than what I had in the video.
Thanks for your videos! Fav midj channel !
Comme toujours tes vidéos sont de véritables pépites ! Très certainement le meilleur abonnement sur l'IA ! Merci pour tous tes conseils et astuces 🙏
Merci d'avoir suivi mon ami ! Je vous tiendrai au courant au fur et à mesure que de nouvelles fonctionnalités seront publiées.
Excellent! Thanks.
Awesome prompt tips Tao brother and appreciate you featuring my inquiry, I hadn’t thought of your suggests so def trying out BIG UPS! Also great techniques for beyond beginners indeed 💥⚡️👏🏻💯
Thanks for your suggestion man! I do hope Midjourney let's us inpaint our own images eventually, that would product photos so much better.
Just watched a bunch of your videos -- so helpful. The big thing I'm working on is trying to create multiple images with the same characters in the same location but with different story beats -- aka make a comic book. Thanks for any help!
Getting the consistent scene is tricky. You could generate a background scene and then use the vary region inpainting tool to add in different characters into that same scene. It is hard to get that same scene from different viewpoints though.
Great videos. Concise. Keep going.
The /describe in the new site is a little sign option on thumbnail images you uploaded., delete and describe as overlays.
Thanks, I see the describe button now. It would have been hard to notice, but it's nice that they included that in there.
Another great video! Thanks for this!
For a future video could give hints on generating images of persons for character reference that doesn't being look odd?
All my attempts end up with persons with eyes widely separated, or extremely oval faces.
You could try specifically prompting for the face shapes you want i.e "square jaw" . I find that Midjourney's portrait photos tend to look pretty good
Thank you Tao! Another helpful video!
Thanks Val! I'll keep you posted on any new tips & tricks.
Can you do a video on create TEXT in MJ? It's constantly creating mispelled words... and how to use the VARY (Region) to fix them - what are the proper prompting editing/adjustments, etc.
It does struggle, especially with longer text phrases. I test it out and see if there's any good ways to deal with it 👍
Thanks for the inputs. Learned a lot for my next videos.
You're welcome! I'll try to make more compilations like this in the future.
Helpful as Always ❤❤❤
I'm always happy to help!
Great video, thanks!
Thank you!
Thank u for ur help and great content😊🙏
I do hope these tips were helpful, I tried to find questions people asked a lot.
One thing that helps me get pics of average looking people is starting the prompt woth “IPhone photo of…” and setting the -s value to 0
Absolutely, I've seen that work also. People take a lot of regular photos from their phones.
Yes, I asked for smartphone photo of people in a specific workplace setting and they turned out like regular, non-model people.
I stopped using LeonardoAi and been only using their Universal upscaler and use Midjourney. When would you use Leonardo AI instead of Midjourney?
Leonardo gives more control of character poses with pose & content reference.
@@taoprompts Thanks for getting back to me.
Hey Tao, do you use an enhance AI like Magnific AI to get better results?
I haven't used Magnific before, I've tried Upscayl though and it works pretty well. It's also free to use which is nice, although the quality may not be as high as Magnific. I've uploaded a tutorial for that here: ua-cam.com/video/BRx9OiBOUWE/v-deo.html
Great channel, keep it up. Could you do a full movie scene in pika? It should now be possible and I'll be trying myself soon to build a short movie for my channel, want to see how our workflow would compare.
I have some videos guides planned for Pika and Runway, I just need to find some time to get them done.
dumb question: do i still use --v 6.0? or no longer necessary?
The version 6 model should be the default now, so there's no need to enter --v 6 anymore.
Thanks
Hello Tao, how are you? Sorry if my translation isn't good. How do I generate images in Mid with 1920x1080 resolution. I tried old commands but they don't seem to work anymore.
You can add --ar 16:9 to the prompt for that aspect ratio and the use upscale on the resulting image. This should produce an image with resolution 2912 x 1632 and you can use any free image editing software to resize that into 1920 x 1080.
@@taoprompts wonderful, thank you! Your videos are spectacular. Thank you for helping us
Hi man .I found your channel on UA-cam ,and want to say thank you for posting this helpful material .But i don’t know why I can’t get result with this method.Every time MJ creates a different person.Maybe I should use discord,exept a web version?
What method are you talking about? Do you mean the new consistent characters (cref) feature. The discord and web version work the same.
@@taoprompts
1-create portrait foto
2-cref, cw 100, style raw
3-mixing different types of shots
and every time the shape of the eyes, lips, eyebrows and other small details changes. Perhaps MJ algorithms are not ideal yet. Maybe stable diffusion works better with characters?
@@tryzem Midjourney will be able to replicate the main characteristics but a lot of the times, the details will not be the same. Stable Diffusion + LoRA or InstantID may work better if you want the specific details down. I'm not sure if those are good with multiple character in one image though.
Hey Tao, do you prefer working on midjourney website or discord?
I'm more comfortable with discord because I've used it for so long, but I think I'll move to the website soon, it is much faster to search and organize your images.
I can't solve a problem for a long time.when I make short videos for example I create a woman picture.a girl with long hair is sitting in the forest.in runwayml in the motion brus tab I mark only the hair of the girl.so I want to move only the hair.I do ganerate but even the girl in the picture moves.i want to move only the hair of the girl.i mean i want to make it like a light breeze of wind.with which application do they do this.i create the pictures in midjourney.the ancient greek girl is sitting in the forest.in the video i want to make only the hair like a light breeze of wind and i do it in runwayml but when i create the video everything moves not only the hair.can you help me with this?
I haven't tried out the motion brush in runway ml yet, it could be that the motion brush just doesn't have that level of fine grained control yet.
thats so cringe when somebody ask "may I ..." and you answer with "yes you may ..." LOOOOL
I just started using mid-journey again after about a year of discontinuing and I must say, I'm incredibly disappointed. It's amazing how little progress it has made over the past year. there seems to be definite problems that it is completely unable to solve.
They have slowed down a bit on feature releases. I think they have had some issues training version 7.
Your education is very good, I wish you could tell it differently. For example, if you could make a famous actor any hero. different different scenes. this is the part I have the most difficulty with. for example arnold schwarzenegger in the role of an ancient greek hero. it would be very good. because although I tried hundreds of times, I could not succeed unfortunately. in one scene Arnold is fighting with a lion. in the other scene the same character is fighting with a dragon. if you make a training video for this, we would be very happy.
Do you mean you mean, it's not able to follow your prompts correctly, so when you ask for Arnold to fight a lion or dragon it's unable to consistently produce the same types of images?
@@taoprompts For example, I depict Arnold as an ancient Greek warrior. For example, when he goes through the forest, he meets a lion and fights it. When I type prpmpu and create the image, sometimes two Arnolds appear, one has the head of a lion, the other image has a lion and Arnold, but Arnold has a tail.
@@aresleoric7455 You could try to use the vary region to and inpaint the part of his body that was generated incorrectly. Like selecting the tail with vary region and then inpainting it with some background. Midjourney does mix things up sometimes though. Usually by reprompting and remixing you can get it to look much better.
That last commentary from Nathalie totally fake 😂😂