Use Midjourney's New Edit/Retexture Like a Pro

[@downrock.creative]

Thanks, man! You're my new go-to for in depth Midjourney tutorials. I really appreciate your trial and error approach to getting individual likenesses. I feel many many AI tutorials focus on the quality of arbitrary results (text to image or text to video), but I'm wanting to see more control over the input and outputs like you demonstrate here. Really well done and well explained!

[@antonpolinski]

Thank you the 1:1 "break the law" was the insight for me! Best MJ Tutorials!

[@glibatree]

Yeah I used to drag the crop handles in closer and closer until I ran into that little quirk. A huge time saver!

[@silversurfer4195]

@@glibatree In the video, you seem to say that clicking on the 1:1 button unlocks the Image Scale beyond 100%. I was puzzled that it wouldn't work for me. I came to the comments to see if there were any relevant and found your answer, which prompted me to drag the canvas handles, and voila! In any case, a wonderful video series. I hadn't been paying attention to your content, but I sure will from now on (again).

[@RussMcClay]

Excellent overview of edit and retexture in the web app. Very cool.

[@TellMeSongs]

Again, very useful and specific actionable advice. Well done!

[@glibatree]

Thanks so much!!

[@poookiepoooo]

great start, will be amazing in 6 months for sure.

[@glibatree]

I'm really thrilled about this feature, it really opens up what we're and to do!

[@resumindo857]

I feel like Generative fill has been doing this for months now, same thing

[@chenfang2331]

Great video! It's great to see MJ joins the editor/canvas competition. MJ's retexture is kinda behind in the game. I tested it and it changes face and race. In my opinion, dzine's image-to-image feature is a lot more mature and offers more control options. I have been using it for 6 months and find nothing comparable to it on the market.

[@gustavo-brunofernando]

Единственное, кто подробно объясняет. Остальные быстро что-то потыкали и мол, ну вы поняли? ну нет так нет. Тут конкретно - да мы сейчас это сделаем и получим результат. Но нет =) Короче, афтар крутой! Показывает сразу ошибки, чтобы предостеречь от них. Не могу сказать что все понял на английском, но сам факт, уверен при пересмотре несколько раз пойму. Спасибо автору.

[@ai.ai.captain]

Thanks for the detailed explanation..subscribed ❤

[@glibatree]

Very kind thank you! I'm very glad you enjoyed 🫡

[@thewebstylist]

Next Level!!🎉

[@glibatree]

This was a really awesome feature to play with

[@goodvibesberlin]

so well explained! thanks a lot =)

[@glibatree]

Glad it was helpful!

[@maxdronov]

i tried to draw on top of the figure - to add some elements under - but this does not work - only changes clothes of the object

[@samsaraAI2025]

Is there any parameter to control how much effect the retexture does? like very slightly changes or really a big change

[@lukewilliams7020]

How comes there’s still no download or favourite button actually on each generated image on the web browser. You have to do several clicks to find those and yet they’re probably the most used.

[@DavidHaydenFreeThinker]

I've been using the retexture feature and the AI moderator has been driving me nuts. Not even trying to upload any images that are in anyway NSFW and not prompts anything NSFW either.

[@Lu-Heringer]

Not available for me 😥😢😭 It's there but I must pay a yearly subscription to use it... Can't wait to have this future available to everyone! Thanks for the video!

[@glibatree]

Hope you get it soon!!

[@JhonataCosmo]

Ideogram has the same feature

[@ThisFrameDoesntExist]

Thx always useful ❤

[@Dose_0x0]

I spent most of yesterday trying to do something similar and ran into the same issue, unfortunately for my use case I needed to retain the face exactly, and so airbrushing it into a 15 year old as with your wizard example wasn’t an option. It would be amazing if you could choose to have the new in-painted part of the image conform to the original lighting, or vice versa have the option to re-light the original part of the image to fit the in-painted lighting, seems so close…

[@glibatree]

Yeah, as far as I know, I took it about as far as midjourney can reliably do. So, not sure what your use case is, but it may be that you will have to photobash your creations in a traditional image editor. They keep improving, but there's still a lot of room to grow

[@armondtanz]

Dang. I watched adobe showcase there new features and it was perfect. Shame about sll the hoops u gotta jump thru. Im not even sure if they will fix them. Its been changing images from the very beginning. Im curious to know if theyve addressed this and if they are going to fix it down the line. Thanks for the tips. Great work as usual 👏👏👏

[@glibatree]

Appreciate you!

[@murphyuriaguereka4360]

Thank you Sir, great video!

[@WhatYouPonder]

Can you do object references??

[@KingZero69]

bro thought he could fool midjourney by putting 'young man'... when midjourney saw right through him and put 'man in his late 30s' 😂😂

[@glibatree]

You got me with an honest to goodness chuckle 😆

[@CodyCha]

He actually refused to read “in his late thirties” 😂

[@glibatree]

@@CodyCha I struggle to hear the truth hahaha

[@zacharyantle7940]

I still don’t have access :(

[@glibatree]

Sad, sorry to hear. I'm sure you'll all get it soon - in the meantime there are some good techniques to learn.

[@JANjeweetzali]

Comment for the algorythm

[@heartrocket]

Do you think Midjourney is becoming "woke"? I generate a lot of images for thumbnails using prompts like "beautiful blonde women smiling," etc. However, I noticed last week that it now produces images of women who don’t fit traditional beauty standards, in fact far from it, including overweight women when I use the term thin. I haven’t seen anyone else discussing this online, but I noticed they're doing some testing in their Discord on what "we" consider beauty.

[@glibatree]

I wouldn't worry too much about the testing they're doing. I believe most of it gets funneled into the personalization models, so if you do some ratings of your own and use --p, I think you'll pretty quickly get what you expect again. But let me know if you notice the trend continuing.

[@heartrocket]

@@glibatree 👍

[@MichaelMarshall53]

That is not the meaning of "woke", the meaning has been diluted so much people do not understand the word anymore. Anyways that seems to be a coherence issue that happens sometimes while training models, but I haven't encountered the issue personally.

[@frankiecooper]

“Woke” doesn’t mean what you think it does, my man. 😂😂😂

[@heartrocket]

@@frankiecooper yes it does

[@JjackVideo]

Finally MJ becomes actual useful. In some months it may be great.

[@glibatree]

Eyes out for v7 👀👀

[@annatyumentseva1248]

+

[@johnovercash1798]

But this didn’t work anyway, even with the reference you got a person completely different than you.

[@OGdigitalpink]

Bro, just character ref with the original image. Easy.

[@glibatree]

Yeah, but the cool thing is this keeps pose too. A pretty exciting application imo, but to each their own. Not everything is for everyone.

[@AndreasAhston]

Steal like a pro.