How to Get the Image You Want in Midjourney: Stylize, Quality, Weight Parameters Explained Visually

[@CorporalDavis]

as an traditional artist, i do believe that there will always be a need for artist... ai is like any tool..such as when the digital pen and digitizer first came out evolving to tablets further on.. into digital painting.. Artist come from use of imagination and ideas to bring into a sort of reality.. which without that no Ai can do! That is a human element!!! Creativity as well as imagination. With that some will use this tool for faster results in whatever medium they choose to use.. Want to say thanks so much for the information and continue such good work !

[@computationaldesign]

Thank you for the comment. I try to decipher & explain every technological achievement/tool in a human-centered way... well, because we have no other option! At the end of the day, it is always about what we see, what we touch, and what we experience. It is good to have this kind of "instant-excitement" generators such as MidJourney on the way though. I expect these kind of tools to become common place in visual concept explorations. This is somewhat the next, and yes, super-powerful visual-making tool. And it can only shine in the hands of visually and soulfully (and syntactically!) trained.

[@computationaldesign]

@@Kcreilly I have a slightly different view on this, let's go back at the time when all drafting in design was made by hand, and when all art was made in analog medium,. When computers emerged, and once modeling, and drafting tools on computers started spreading, people started discussing this : "Can we consider the art created on computers as real art?" Of course, looking back at that discussion today, we find it funny. But there are many milestones in human history at which a tech was considered to replace many values in human life or render humans totally unnecessary. Maybe a pessimistic idea, but ever more so an inaccurate prediction. AI is not a standalone thing on its own, in MidJourney's case we are dealing with specific algorithms that are quite good in semantics (deciphering what a word, or sentence means) and how a prompt could be rendered by using pre-learned patches of pixels and patterns. An artwork's value is determined by the one who pays for it (beauty is in the eye of the beholder -- and I am skipping the politics and populist motivations here). Whether a computer generated art could be mixed with a human artist is not a real problem in this respect. Let's also keep in mind that AI is also a "real" work humans created as well. It is not truly replicating the human elements of creativity as it has a completely different mechanism in generating images. So I'd say it is an extended period in human history (AI research started in around mid 1950's, and soon we were looking at computer-generated art). Whether we use these technologies to generate happier lives with creative outlets and sustainable environments will be up to us. Let's not make it a mistake.

[@RabidDisposition]

@@TerraStory225MYA this is the sad reality of it. In probably less than 5 years, this tech will be good enough for companies to completely forgo hiring graphic designers and concept artists. I already know a company owner that generated his company's new logo in midjourny, saving him literally hundreds of dollars. No on is gonna want to hire some human with an inflated sense of how much their art is worth.

[@gerrycanyerryyerry9641]

AI art is such a buzzkill. I hate it. Sure it's a tool, but I wasted my life for nothing.

[@RabidDisposition]

@@gerrycanyerryyerry9641 to be fair, I'm sure you've been told that art would pay the bills before. You made your choices

[@netpark7025]

TNice tutorials was a very informative video. You made soft soft soft one hundred tis easier! Thanks alot

[@computationaldesign]

Great to hear!

[@jasantana]

Onur!!! small world. Glad to see your teaching this topic, I'm super hyped for a.i. generated art and I been having a blast using Midjourney to create art for my content. Keep up the good work

[@computationaldesign]

Small World indeed! Glad if this was helpful!

[@221Mono]

Thank you for the summary! I've been experimenting a lot too, but looking this video really helped me to sort the aspects and understand the parameters better. Very nice overview.

[@computationaldesign]

Awesome. So happy that this has been useful!

[@DopelyTV]

agreed

[@computationaldesign]

@@DopelyTV thank you!

[@KokohNicky]

Thank you. I really appreciate the flowchart. I find it very helpful, and certainly generous of you.

[@computationaldesign]

Glad it was helpful! I have the PDF link in the description if you are interested.

[@Splinterverse]

I'm using the PDF. Super helpful video and PDF. Many thanks!

[@computationaldesign]

Great to hear that the content was helpful.

[@skryonix357]

Congrats to David and the team man! I’m glad they’ve gotten bigger than even the Minecraft video game server, which discord was initially made for.. gaming. Midjourney is such a beautiful tool.

[@computationaldesign]

I learned something today coupling. Thanks.

[@DopelyTV]

thanks for the info bro!

[@computationaldesign]

You bet!

[@Gprolico]

great content man! thanks for sharing

[@computationaldesign]

My pleasure!

[@БлаблаБлаблов]

Thanks! Clear explanation!

[@computationaldesign]

Glad it was helpful!

[@aldodibari]

Hello, excellent tutorial. Do you know the manual that describes the parameters and their effects in creating an image in MidJourney? Thanks.

[@computationaldesign]

Yes, which are informative but not comprehensive, and I think the community has been sharing lots of developing by making similar experiments to the one I am sharing here. Thanks for the support!

[@robfielding100]

Very helpful video. Thanks

[@computationaldesign]

Glad it was helpful!

[@traviscislo6260]

--quality actually just extends the time of the diffusion process allowing more details to be rendered over time. --q 2 = 2 minute render , --q 5 = 5 minute render

[@computationaldesign]

Great point, and again --quality is currently referring to the [duration of the generative process] whereas "quality" has a different meaning in visual arts and even a different one in digital production. All in all, I am thinking --q could be rethought to have another parameter name. --q5 was removed a while ago.

[@traviscislo6260]

@@computationaldesign yes I like the idea of naming them something more relevant. Simple things like descriptive words are surprisingly influential in the process of art. Btw, -q 5 is back. Try it out 😉

[@traviscislo6260]

@@computationaldesign Have you had the chance to do studies on -chaos yet? I still can’t figure that one out. It looks like it is adding extra parameters with randomized values depending on the value you originally attribute to it. Varying colors and even subject matter sometimes. Maybe it adds random input images? Either way I get very different results each time I use it so nothing consistent enough for me to theorize with. I guess that makes sense considering it’s labeled “chaos” ☺️

[@computationaldesign]

@@traviscislo6260 now I am afraid that I would eat all my fast time with --q 5!

[@computationaldesign]

@travis cislo Head on! Actually I did play and have a similar feeling. I have a collection of images again to explain it but my overall sentiment is very close to yours [nothing consistent enough for me to theorize with]... I'd say if you keep experimenting you discover more about the affect of the --chaos parameter. Yet all of this is experimental and joining the MidJourney community calls I saw that the developers are also experimenting along with the community, and nothing is really "hard-coded". As you suggest this is all natural as we are using language and ambiguities are to remain...

[@ChrisBamborough]

Great video, thanks for making. I’m fascinated by this topic, but I’m unsure what the value of the output is. What do you use the generative image for, is it an end product or is it an inspiration for other art forms?

[@computationaldesign]

If we take it as an endproduct, then we end up with too many end products! I constantly work on developing design workflows which help filter out the clutter, bring forth the good stuff, and use those for getting tangible results.

[@phgduo3256]

Thank you so much! What a great explanation.

[@computationaldesign]

Thank you for the feedback. My pleasure.

[@Maubro1]

Great! Thank you for the tutorial :)

[@computationaldesign]

Thanks for watcing!

[@fundoozzz]

Thanks for this excellent video, helps a lot. One thing I wanted to know is that is there a parameter to control the size of the car as compared to that of the complete image?

[@computationaldesign]

As of now I don't think there is a parameter to do this, but it is possible to do it using signifiers in the promt. Try using 'full' 'three quarters view' 'in perspective' etc. In the prompt. Describe what you want to see... one should work.

[@kimmiekardesh8514]

BRO! upgrade your mic bro! thanks for the info though

[@computationaldesign]

Yes, had a glitch with sound recording in this one. Thanks for the heads-up.

[@EMY.sr.]

Teşekkür ederim

[@computationaldesign]

Ne demek...

[@memoriespast1499]

But how do I revert back to the normal settings

[@computationaldesign]

You can use /settings to set global parameters. Any local (per run) parameter will be defaulted if you don't use them at all. For instance instead of using --s 625 or --s 5000. if you don't use --s parameter at all, the default parameter (2500) will be used.

[@Gundesalf]

Here's a challenge. Make MidJourney create the image of a sports car that, instead of wheels, has feline feet. I tried different contexts and modifier at no avail. If someone manages to achieve it, straight out of MidJourney (no post production or Dall-e replacements), please post here. I'd like to see how you did it (the text that finally spawned it).

[@computationaldesign]

Good challenge, I got some results @Heimrich Gundesalf but I'd say they are less than 50% successful. I can think of better ways of testing this, but need to make time to test.

[@abdullahsy]

Amazing details and ground up explanation how it all works, I really appreciate it. I have to say something that makes me worried a bit, I've been using Midjourney for some of the projects I'm working on, the ones that are apstract and wild, lately I started to realize how much it is growing on me, it's easy and original in a way, quick and detailed at some point for a quick draft. I'm worried though about myself, if one uses Midjourney or Dall-e or any other Ai out there, during the initial phases of any given project, 🔴 Does Ai inhances our imagination or is that just the hallucination we get before we turn totally Cyborgs and addicted to Ai? 🔴 And/or do we lose our capabilities of self imagining and self constructing an idea in our head without the use of such tools? Thank you!

[@computationaldesign]

No need to worry-- if it is growing on you, it means you are visually driven. I agree that this tool is unlike nothing else for getting well-"crafted" conceptual images (along with other text-to-image tools). Quick answers to your two questions are: no & no. It is natural to feel uneasy when a paradigm-shifting tech appears, or becomes common place. These technologies existed for many years now and only now they are becoming more accessible. We can see similar leaps in the history of design, manufacturing, transpiration and communication. Namely first industrial revolution, and then passenger planes, and then internet, and then finally 3D printers, etc became commonplace and they scared some people while they encouraged others to make false predictions about the future. Longer answers for your questions: 1- Text-to-Image tools are using natural language processing along with some parameters to generate images, and it is, yes, just another tool that help us realize our visual ideations. 2- The only bad thing that may appear here is people becoming a little lazy: to generate similar images, you'd need to spend many hours modeling, collaging and painting in the past. You just type the words and the image appears now. Yet this also happened in the history of the civilization, tools, gadgets and technology made us a little lazier. The cool thing is you can leverage this recurrent event in the civilization and instead of becoming lazy, you can spend the spared time for growth: How can you tame MidJourney? How can you learn the inner workings of the tool? How can you use it in a more meaningful way? Can this tool help you develop your visual, syntactical and intellectual skills? ...and so on. As long as our heads are connected to our bodies, and our bodies keep exploring and expriencing our environments, we'll keep imagining and conceptualizing. I hope this helps--

[@mattterrones8999]

Hello. Wonderful video. I subscribed. Please share the PDF! Thank you!

[@computationaldesign]

The PDF is available in the description now. Thank you for your interest.

[@troncalnegrete479]

Thanks!!

[@computationaldesign]

You bet!

[@DRIVECLUBistimelessPS4]

are you planning to get a better mic for your videos? :D

[@computationaldesign]

The cause for the glitch was the recording interface I was using. Sorry about it!...

[@untitled795]

Very helpful, I had a question (and maybe i missed the answer in the video), what is the slide app or website you use to display the matrix? Thank you!

[@computationaldesign]

I tracked and prepared the slides manually: meaning no automated process there. I prepared the matrix in Adobe InDesign. You could use any tool, google slides, etc would work of course. It would be lovely to have this kind of lineage matrix as a feature in MidJourney of course.

[@mar_ink]

check out "Miro" for this type of things!

[@jungsomewhat]

Perfect thanks

[@computationaldesign]

You're welcome!

[@sikliztailbunch]

It would make much more sense to use a fix seed for tests like this

[@computationaldesign]

Would you like to explain why so the community can benefit from your comment? And you are partially right and partially off, could you guess why? (If I may rephrase: "using a fixed seed for such a test COULD ALSO make sense, for a more convergent comparison, or a more strict controlled experiment.")

[@matoxetu]

very informative but the audio quality hurts

[@computationaldesign]

Yes, sorry, my bad with the audio in this one.

[@1deplatt]

How do you add more variation to the variations??

[@computationaldesign]

Use the V1 - V4 buttons to make 4 variations of each variation.

[@mrschneebly85]

The Sound!!

[@computationaldesign]

Noted. Thanks!!

[@LuigiCotocea]

It uses Vqgan or Disco diffusion?

[@computationaldesign]

MidJourney!

[@r.b.667]

great... it would be even better, if the chart will be available for download...

[@computationaldesign]

It is available, check the link in description!

[@r.b.667]

@@computationaldesign thank you for pointing me to that. R.

[@iampreppietrash]

subscribed to channel and liked video, I sent a message for share request.

[@iampreppietrash]

great idea, makes total sense to have a cohesive and repeatable workflow for experimentation. Thank you!

[@computationaldesign]

@@iampreppietrash Sent the PDF right now. Cheers!

[@MadelineMysterious]

wish people were more hard focused on AI for health issues not this kind of thing. My mom works is a lot of AI these days and most of it is to cash in like the midjourney developers and not solve any relevant problems.

[@computationaldesign]

I hear you, but doing some research may sooth your feelings: 1- check the AI in healthcare market which is already huge and expected to grow 10 fold in next 6 years. 2- check the past projects in distributed computing, and how people were engaged with it (folding @ home, a project I also contributed to for around a decade) 3- midjourney has been self-funded so far. no investors, no venture capitalists are involved. join their community calls if you want to learn more. 4- last but not the least, one matter or technology is never meant to be used for one single purpose only. Think water: we drink it, we use it for hydroelectricity, for washing our cars (somewhat a waste), for swimming in it, etc. Think petroleum: used for transportation, for generating heat. Think about computers: used for solving many problems, yet so many play games on it... If gaming is not solving any relevant problem (?! pun intended), why do people do it? -- so then, for well-being humans need to be healthy, but also need to flourish their lives with literature and visual arts. Midjourney is helping people play in the latter, and cannot be held responsible for the things that reside outside its scope. I hope these help.

[@ethnicalbert]

dude please get a mic. thank you

[@computationaldesign]

noted!

[@thevikingsock8527]

dude, get a new mic, the sound quality is awful. Its not expensive

[@computationaldesign]

Yes. my bad with recording settings, otherwise the dude has mics.

[@Maidaseu]

"takes more active effort than gaming"...lol my friends and I use to create calculators, PCs, artworks, cities, architecture etc in Minecraft many projects requiring 100s of hours.

[@computationaldesign]

"*Creative*" not "active." I am referring to visual creativity, that requires a trained eye, knowledge about artistic styles and the challenge of developing those in midjourney, using natural language, I.e. semantics. This all relates to handling aesthetics in computational terms. There is a big difference between *labor* and *creative* (mental & perceptual) effort. I am talking about the latter. Quantity doesn't mean quality, the hours we put towards something does not directly correlate to its quality-- not to demean your effort, but to clarify the context I am trying to build. It is impossible to talk about these in couple of minutes I spared here for the news. And I am not sure if you would be interested in checking some of my other videos to hear more as you gave me your "LOL". Nevertheless I can understand, and actually know about the amount of work you are referring to. Keep pushing, and keep being both *creative* and *active*. Cheers!

[@Schaudochnochmal]

FAKE IT TILL YOU MAKE IT!

[@computationaldesign]

Works for some...

[@sebastiannazarevscky3496]

amateurish, non scripted rambling, horrible audio! in a word, lazy! 30 min could've been 8 min at most...

[@computationaldesign]

Dear Sebastian, Good feedback (I liked it), but could have been better. Let's analyze together and understand why. I itemized your feedback and made a quick assessment that shows how accurate they are, and how much of them I could use to improve my content. See your score for accuracy (I will try to be objective but, you know, we are all subjective creatures, so please keep that in mind): 1- amateurish (50%) - yes for KZbin, but not for the quality of material 2- non scripted rambling (25%) - you cannot have a high quality PDF w/o a script, yet the delivery could have been better 3- horrible audio (75%) - yes, a setup mistake on audio recording interface caused this, but still listenable 4- lazy (0%) -- this is amusing for me as you are talking about something you have no idea about. 5- 30 min could have been 8 minutes (35%) - maybe 15. Not 8. 6- Delivery method and tone (1%) - this is where you have the biggest room for improvement, as far as I can see. So your overall score for this comment turns out to be around 31%. I will still take it as I want to do better... But if you want people to listen to you, you may consider working on the missing %69. There are a lot of resources online on giving feedback. I would encourage you to skip this video (or my channel completely) and better use your time on something that will help you truly grow. I hope you make it one way or another. Good luck-- (p.s. This is another way of giving feedback.)

[@PuffyNavel]

@@computationaldesign You've got a great attitude! An attitude that guarantees your content will improve, automatic sub for me. Yes, audio is bad and you acknowledged it✌🏻Have only watched a couple minutes and there is some rambling but I feel confident there's some good, to the point, information in here. Looking forward to more educational content!

[@PuffyNavel]

@Sebastian Nazarevscky What’s more lazy? Giving it a go and sharing your knowledge (and inevitably meeting some bumps in the road) whilst moving towards improvement. Or writing an impulsive, toxic, entitled comment on a video of someone trying to help others while you yourself have no videos on your channel. What's your contribution? What are you adding to the world by leaving trash comments? Look inwards and watch where your comment stems from. Did you have a purpose with it, or were you just lashing out on someone you have no reason to feel so indignant towards?

[@computationaldesign]

@@PuffyNavel Thanks a lot for the support!

[@ent4ever30]

@@computationaldesign bro leave negative people with negative comments. You are doing great keep it up buddy actually for me this video was small I want more.

[@aldodibari]

Hello, excellent tutorial. Do you know the manual that describes the parameters and their effects in creating an image in MidJourney? Thanks.

[@computationaldesign]

Yes, which are informative but not comprehensive, and I think the community has been sharing lots of developing by making similar experiments to the one I am sharing here. Thanks for the support!