The Good, the Bad & the Ugly of Midjourney INPAINTING

📙 Midjourney COURSE mastersofmidjourney.com Beginners to Advanced
🚀 FREE Midjourney Cheat Sheet tokenizedhq.com/freebies/mj-c...
🔗 FREE Supporting Material tokenizedhq.com/freebies/vide... Folder: 2023-09-02 - Midjourney Inpainting
🔗 FREE Promptalot Extension promptalot.com/extension
-
📰 AI Newsletter 👉 Your Inbox tokenizedhq.com/newsletters/ai
🐤 Follow me on Twitter / chrisheidorn
💼 Follow me on LinkedIn / christianheidorn
📸 Follow me on Instagram / christianheidorn
💬 Join the Tokenized AI Discord tokenizedhq.com/invite/discord
-
📺 Recommended Related Videos & Playlists 📺
Commands, Syntax & Parameters: • Commands, Syntax & Par...
Prompting Methods & Experiments: • Prompting Methods & Ex...
Character Design Series: • Character Design with ...
Storytelling: • Storytelling with Midj...
Watch Me Prompt: • Watch Me Prompt in Mid...
This Brand Does Not Exist: • This Brand Does Not Exist
Midjourney for Merch: • Make Money with Midjou...
-
Midjourney's recent unveiling of their interpretation of inpainting has the tech world buzzing.
But is it all it's cracked up to be? After intensive testing, here's the lowdown.
The new feature, "Vary (Region)," offers a spectrum of use cases from adding or removing objects in an image to adjusting facial expressions.
It even allows for a degree of face swapping. But while it excels in some areas, it also falls short in others.
Replacing entire characters in a scene, for instance, can produce odd results.
And sometimes, the system seems to misinterpret prompts, leading to outputs that don't align with user expectations.
The bottom line? While there's significant potential, and some results are downright impressive, there's still room for improvement.
Dive into this video for a detailed walkthrough of the good, the bad, and the outright ugly of Midjourney's newest feature.
-
⏰ Timestamps ⏰
00:00 The Good, the Bad, & the Ugly of Vary (Region)
00:27 Removing & Adding Objects
04:56 Removing Objects in Condensed Environments
08:41 Replacing Scenic Elements
13:19 Face Swapping
15:51 Facial Expressions
17:16 Composing Outfits
20:30 Replacing Entire Characters
22:31 The UGLY
#midjourney #midjourneyv5.2 #inpainting #varyregion #midjourneyinpainting

[@manfrombkk]

Yes, for face swapping, expression changes, and of course outfit swapping,, it’s a Total Win. Combine it with Photoshop’s Beta, generative fill, and you can move your character mods to a new level of creativity.

[@johnfromberkeley]

Great video, as always! BTW, it’s an that “pillow” is an ottoman. 😂

[@TokenizedAI]

Ha, thank you! 😁

[@mygicarskrsk4465]

thanks for the amazing value you bring!

[@user-en3cs9kq5g]

Christian, great video thank you, I learn so much form you. :-)

[@ScaryStoriesNYC]

I've been using it to create werewolf transformations for my channel. It works well about as often as any other inpainting IMHO. I get as much done in MidJourney as I can before trying other things to fix the rest like clipdrop and photoshop, whichever seems like it might do a better job going by a gut instinct.

[@berengereu]

Personally, I think it works great. And in the case you asked for alps + snowflakes, it worked correctly somehow, because you cannot have that light interior and have snowflakes and clouds (dark sky) at the same time. So I am happy that it does it like this, I want MJ to ensure the image remains coherent (especially in terms of lighting). For putting a female detective, maybe giving a reference image would have helped, and maybe worked.

[@user-dq1qq1ut3l]

thought the same thing on the snowflakes. this is why it is important to have some type of discipline in the visual arts to get the absolute best out of these AI tools - a talented painter, photographer or cinematographer would have picked up on lighting of the scene immediately.

[@puja1985]

I'm glad you took the time to study the inpainting feature before making a tutorial on it. A lot of KZbinrs jumped on the bandwagon and made tutorials on the very first day it was released, which resulted in incomplete reviews.

thanks !!!

[@antidiangelita]

thanks for share,, for me ,, is a great help

[@travelpointone991]

Regarding Turbo mode, Midjourney notes that if they are no servers available for Turbo it will use Fast.

[@timothypruitt4578]

Good video as usual. Using the techniques you demonstrated here I was able to make corrections to a character I've been attempting to design for months, attempts that have been met with mostly failure. Today, the changes I needed to make were a breeze.

[@harrygoldhagen2732]

Very useful, Chris, thanks! I especially appreciate your showing the "ugly" -- which is so often my experience in MJ! Anyway, the new inpainting does look helpful for face swapping, clothes swapping, and maybe even fixing bad hands and arms. Looking forward to seeing further experiments.

[@kurdonoid]

Thanks! figured I'll stick with SD for now, and photoshop plug-ins!

[@SpiritTracker7]

Try “it’s snowing’ instead of snow flakes. We never say it’s snow flaking outside. Could be confusing the AI

[@AG_before]

Thank you. 👍 I'm actually surprised about it fixing the hands. I'm not sure why that hasn't worked for me - perhaps I should just prompt 'hands' in the remix. But overall, it's been a miss, miss, miss, possible hit, wth, miss, hit, miss journey for me. Considering MJ is already a bit of a beautiful slot machine, it's an unfinished feature. And I believe that's what they wanted! Appease, beta test, and eventually, please with a polished version. They're certainly business savvy. :)

[@DigitalAscensionArt]

Thanks for making this video Christian! Any success with just inpainting body parts like replacing a bad hand for a good one?

[@tomi71]

Thank you! Very interesting as your videos always are. I wonder if we can make the character look into a different direction with this (or maybe it is possible with just using remix. Can´t test as I don´t have it right now.) Just out of a little curiosity, was this a bit faster than normal. Like 1.2 or something. It made the video shorter which I suppose is good thing on this short attention centered world. Great video. Thank you.

[@timothypruitt4578]

The Ugly looks like a Star Trek transporter accident.

[@Dazreil]

try image prompting with an image you want to add

[@TokenizedAI]

Though I really shouldn't have to do that. 😅

[@TeddyLeppard]

This proves that Adobe is on the right path versus the results possible here.

[@MONTY-YTNOM]

Its called a Pouffe

[@GillBrooks]

Another interesting video and more options to try. Mine has been very hit and miss with this so far. Oh, and that piece of furniture, depending on the style, it's a pouffe, footstool or ottoman :)

[@steveballantyne]

Christian, thanks so much for taking the time to put together these informative videos. I run a small brand consultancy in NZ and wondered if you had any thoughts on what business categories (in your experience) where most receptive to using AI generated imagery?

[@amanaahphotos]

Thanks for this (: A client commissioned me to use this feature to change specific facial features, like make eyebrows thicker, nose smaller etc.. I was wondering if you’d try anything like that?

[@user-tp8il8sd3b]

hello, tu suppress part of the image, you can enter: empty

[@davidmichaelcomfort]

Great video, esp how one can change clothing. I had the same experience trying to place a character in a scene. It doesn’t work that well at all. I was able to get decent results if I placed a character in a scene and got a half-way decent result. Typically the character will not be in the same style as the rest of the image and looks out of place. But using Vary Strong, I can make the character into the roughly the same style. That said, I think face-swapping characters that are already in the image is a better method. However, your results may not be that good if the faces being swapped are not the same size, as you noted

[@morpheus2573]

TY. Great to learn that you can use image prompting with Vary Region for face swapping. When I tried to prompt new faces using Vary Region, without a reference image, it went a bit Stephen King on me. Looking at some of the results from the Vary Region option, it almost feels like MJ has pick-whipped itself to the "-- weird" parameter.

[@blgfxstudios]

it's called an ottomon

[@TokenizedAI]

Thanks 🙌🏻

[@5XM-Film]

Wiedergabe geschwindigkeit normal… hast Du Deine Sprache angespeedet ? Du sprichst so gehetzt 😉

[@TokenizedAI]

Ich wollte unter 30 Minuten bleiben. Hab nur um 1.25x beschleunigt, weil ich eigentlich meist recht langsam rede. Am Anfang merkt man es mehr als später.

[@IridiumZero]

Yeah, it's not really usable right now outside of some very limited cases.

[@williamkaizen4479]

I think because you kept the same prompt . Maybe try to just input what you want . Example : “ female detective “ . Instead of whole things , it seem confusing to midjourney somehow ?

[@Fragsu]

Did you artificially speedup the video? Also, your voice sounds a bit robotic and distorted. No need to use AI for everything you do ;)

[@WilliamAlanWebb]

Did you speed this video up, or drink too much coffee?

[@TokenizedAI]

Maybe both? 😂

[@sassidiego1274]

dude, did you sped up the video?

[@TokenizedAI]

Haha, yeah. Cause I wanted to get the video under 30mins. Turns out it wasn't such a good idea. Gotta experiment though 😂

[@sassidiego1274]

@@TokenizedAI well, I personally liked it, your voice wasn't much altered and there was no wasted time. Furtunately i can understand english pretty well and didn't miss a thing. Keep It like that, i think it's not a bad idea!

[@pushismano77]

Photoshop Beta does this way more better...

[@artcgix]

The thing is that you did get a woman, but maybe she identifes as a man? Great videos, keep it up!