Thank you so much for the kind words. You have no idea how happy this makes me. I think my objective here on TH-cam is different from other channels. I want people to come back because they actually learn how something works...in depth. Not just a superficial introduction to some features. I still have a long way to go in terms of production quality of my videos. But I'm trying to improve day by day.
Best youtuber out there on Midjourney. I only got more confused the more I watched others. Christian makes this now much more approachable. I'm trying to find a way to upload architectural models and have them render landscaping and scene without altering my model too much. Haven't figured that out yet, but I see people upload portraits of themselves and generate what still looks like them, so I figure I can do that with a house model.
I appreciate the effort, and I find this very useful. I've been using MJ for a while now, and although there are numerous text based tutorials, my mind has never learned efficiently in this way. Having someone instruct me slowly, with consistent examples along the way is the exact way I learn everything (unfortunately). Thank You.
Deine Videos sorgen dafür, dass ich immer wieder zurückkomme! Sie sind sehr professionell und informativ. Außerdem ist dein Englisch kaum von einem Muttersprachler zu unterscheiden. Ich freue mich auf zukünftige Inhalte!
Hi Christian, I just watched your video about Seed and Sameseed, as a newbie in Mid journey, your content is Gold. Thank you so much, you've just earned yourself a new subscriber. ❤️
Thank you for your great, calm and in-depth tutorials. I also really appreciate the lack of background music during the tutorial itself. For me, background music almost always interferes with my ability to focus on what is being taught. Of course, this is personal preference, but I appreciate it nonetheless.
My first few videos had music. I think because we're conditioned to think that videos must have music. But I guess it largely depends on what sort of videos you're making.
Loving the videos, You've improved the images I generate by a massive amount! One of the clearest and easiest to follow video series I've found on AI image generation
You are a brilliant teacher. You’ve helped massively with how I’m able to create stuff much closer to how I imagine it. Thank you so much. Liked subscribed saved and shared your stuff 👊👊👊
Oh man I am happy to thank you for the tremendous effort, I do not know how to return the favor, subscription and admiration is the least thing that can be done for you.
Thanks so much for confirming my assumption that a lot of people would prefer more detailed and nuanced explanations. Have had some rude comments recently with people complaining that they've been too long or that I talk too slow. LOL
greetings from germany - great work! AND even with my likely basic english I do understand the main part and what you explain. So I would like to thank you very much to teach newbies like me. So funny that you explained that you pronounce(?) "parameter" in a special way. I do copy your style of "parameter" now ;-) because you are a native speaker
Grüße zurück aus HH ;) Ich bin "offiziell" kein Native Speaker. Allerdings bin ich in Kanada aufgewachsen, weswegen ich mehr oder weniger trotzdem dazugehöre ;)
This was an interesting video. I have to say I don’t really have much success using “no” as an editing command. It’s much more for the original composition
Just like with regular prompts, you need to detach yourself from how YOU would describe something. Instead, you need to think about how someone else or many other people might describe what you are trying to do or trying to remove. Also experiment with different words for the same thing. Most people tend to limit themselves by assuming that the AI knows exactly what they mean. When in fact, our description just might not be explicit enough.
My compliments on your excellent videos. One question, I work in black and white, and sometimes I am tormented by mid-journey's addition of grayscale. Is there any way to minimize grayscale?
@@TokenizedAI Looks like it's my computer editing my words again. What I meant was I wish there was a feature whereby you could edit specific areas of an image, reliably, without affecting the rest of the image.
Hallo Christian, wie stehst du zum Thema Midjourney auf Amazon Merch on Demand? Ich nutze es bereits aber viele haben Bedenken. Ich sehe eigentlich keine Probleme unter Beachtung der Copyrights, welche ich bei Elementen von Vexels auch beachte. (keine VW Bully etc.)
Was genau sollen die Bedenken sein? Die Lizenz zwsichen dir und Midjourney ist eindeutig. Ich wüsste auch nicht, wer konkret Rechte beanspruchen sollte, zumal kein Beweis erbracht werden kann. Wovon ich Abstand nehmen würde, ist die Verwendung der Namen konkreter Künstler. Da bewegt man sich tatsächlich auf dünnem Eis, sofern der Stil ein bekanntes Markenzeichen des Künstlers ist und dies auch ohne AI zu einem Rechtsanspruch führen würde.
What does weight 0 and -1 and below do? do they work? ... what if I use weight 2 or 3 on all elements? is it the same with every number as all elements are weighted the same?
Does it matter if you do the varitiation / remix on a thumbnail BEFORE upscaling it? Will that effect the final quality if you variated an image and then upscaled it VS if you upscaled a thumbnail, and then remix that upscale with the same text prompt in both cases?
Is there a way to make Midjourney consider a 2 word, word as 1 word? I am working on a project where the main character have blue hair and golden eyes, and I normally use a reference image. Where the blue color of the hair tens to bleed into the eyes. If I use the "--no blue eyes". So do it look like it interprets is as "no blue and eyes" making the image a lot more red and still getting the eye color wrong. Also specifying golden, amber or yellow eyes, seems to have no effect. Since seems to consider it as "golden and eyes"
I tend to agree. If you want to change something very specific, you're better off using Inpainting in DALL-E to make the adjustments. And then use an AI upscaler to get a bigger version of it.
Thanks for the tutorials! I found that --no can sometimes warp the entire output. I was trying to get a boy and his grandfather in cartoon style. It usually generated the grandfather bald and bearded, which is not what I wanted. However using --no beard and --no bald pretty much limited me to always the same exact grandfather and style regardless of changing the style, other keywords or seed. It almost feels like Midjourney considers those features an essential part of being a grandfather? Perhaps using a mild negative weight would work better? I tried different positive words (e.g. clean-shaven) but that had little effect. I ended up completely rewriting and specifying an old man with specific features and dropping the "grandfather" keyword entirely.
Using --no is tricky, for sure! However, negative weights are all the same. Makes no difference whether you use -1 or -0.5. You're asking it to remove something, not just remove it a little bit.
@@TokenizedAI Hmm, it almost sounds like --no is best used when the thing you want removed is almost orthogonal to the main prompt. If it's at all related to the main concept MJ will just tie itself in knots.
hi! many times when i create something it comes with some text or similar to text, many times when i want a logo or a poster midjourney adds some text (or looks like text) HOW TO SAY to midjourney no to add any text? thanks!!!!
Unfortunately it's currently almost impossible to prevent Midjourney from adding text elements. Some have tried "--no text" or "--typography" but that doesn't work because the AI model probably doesn't even know that it's adding "text". It can't remove what it doesn't know. The only real solution right now is to edit the image in Adobe Photoshop or another application and remove the text by hand.
I haven't found that using negative prompts (or even weights) consistently work, especially in regards to signatures and watermarks. Even clicking on v1, v2, etc. for whichever one you want variations on and then editing your prompt to either add the negative prompts or using text weights before resubmitting doesn't always work either. And my prompts are never really complicated because I'm just not that smart. (People shouldn't be correcting you on your English since it's not your native language and most American words come from other countries anyway so in your native tongue you may be pronouncing "parameter" correctly...I swear people need to have better things to do rather than be grammar/language police since at the end of the day it really doesn't matter...I love your accent).
You're right, they're rather unreliable. However, I've found it to be helpful if you use various synonyms for whatever you are trying to remove. Sometimes even words that aren't directly related. So for example instead of "--no signatures, watermarks" you could also try adding "artefacts" to the list and many more. The more I exclude, the better the results tend to be for me.
Hi can you give some advice for how to prevent --ar 2:3 streching a figure's proportion? They all look weridly elongated, or at least a tall supermodel. midjourney will do this for 2:3 all the time, I've tried --no streched proportion --no elongated , but it won't work.
@@TokenizedAI yes, it's v4 on default. It's a very common problem, all the other aspect ratio won't have such issue, but 2:3 will produce many elongated characters.
Unrelated, but I'm having such a tough time getting mj to create my character. It's a creature that has a human body and octopus head (kind of like Davey Jones in Pirates of the Caribbean). Can you help?
Sounds like a major headache because MJ tends to blend more than we want to. However, you're actually trying to do just that and also have a very specific idea too.
@@TokenizedAI you’re right. MJ tends to either ignore the octopus head instruction or give me an amphibious texture to the body. Do you have any suggestions?
Is there a way to use a negative prompt where you multiple words? As in “--no small dog” to my understanding Midjourney will just take that as no “small” and also no “dog” as two separate negative weights rather than a complete singular prompt
@@TokenizedAI thanks for the reply, would this work where you are expressing multiple negative prompts I.E -"-no clown nose glasses" in say a situation you have a character you've generated that ends up having a clown nose and glasses, that you want to eliminate, would it not likely take the nose off the character? rather than eliminate the type of nose. I hope that makes sense. From some past previous experience using multiple words in the negative prompt that might describe a singular thing, it sometimes breaks them into separate ideas.
I think so, but why not just do it with a miltiprompt segment. Then it's 100% guaranteed to be interpreted as a negative prompt (not saying it will give you your desired outcome, but the input would at least be correct). "your primary prompt::1 your negative prompt::-0.5"
I have tried using negative prompting for words/text but it does not seem to work. I am creating an album cover and it insists on adding text to the cover.. Any ideas how to prevent this?
Midjourney is an image AI model, not a typography AI model. It's not capable of generating sensibel text. Whenever you see something in an image that resembles text, then it's only because the images Midjourney was trained on, often contain text. The AI doesn't even know that it is creating text-like content.
How do you know the syntax specifically. So negative is - Why is -- not a positive or is it dash - . If that makes sense. Dash dash -- or colon colon :: isn't intuitive to me
It's not math. It's software engineering. They are very common concepts in that industry, because this how many command-line terminals work. I know the syntax because there's an official documentation that explains all this in detail.
I'm pretty sure --no also recognizes spaces. It basically takes everything until it sees another parameter. Alternatively, just use negative weights. picture frame::-0.5
great info but please , level your audio please, I had to jump on the volument to turn is ( so ) down during your extra loud intro , then I have had to pull the volume back up to hear back your voice.
I find Remix so unreliable, that it’s not worth using. Even making small changes to a Remixed prompt often causes the resulting images to be very weird (like your birthday cake example). I almost always just input the prompt again, adding my changes. CMD Z (if you just want to bring up the last prompt used) or Copy Paste if it’s an older prompt.
I agree that it's a tricky one. I'm not sure I'd say it's unreliable. I think the problem is rather that we largely still don't fully understand what it does under the hood. From my experience, Remix will stick extremely close to the composition and colors of the image you're remixing, irrespective of the prompt you enter. My guess is that it uses the image as an image prompt and assigns a very high image weight to the prompt, effectively rendering your text prompt adjustments less relevant. I'll have to experiment a little more until I fully understand how it works.
@@TokenizedAI When remixing do you always keep previous prompt in pop-up window? Do you know if there is any impact on the result when you either keep the prompt or just remove previous prompt and type in new words/commands?
So, don't take my word for it, but I believe that when you Remix, it basically uses the image as an image prompt with a very high weight and then uses the text prompt to create something that fits exactly into the composition, shape and colors of the image. That's why it's pretty unreliable. I recommend trying both. Do extreme stuff to figure out what it's doing.
@@DennisHellberg Funny thing is, it doesn't do that on my system. Yes, it gets louder, but not to the point of being obscenely loud. I think it's just an issue of bad (or inexperienced) sound editing and the fact that everyone has completely different sound setups and default volume settings.
Hi, thanks for the content. Just a suggestion, you should remove the mouth sounds between words from your videos. This will improve the production value and be more pleasing to listen to.
hey guy - come to the point - just a hint for your vids - make it short - and focus on the information to want transfers - hot helpful I would say - so thumbs down
lol, no I wasn't. I was merely illustrating the difference between 3 and 4. Should be obvious enough from the video. 3 weeks ago v3 wasn't set as the default yet. So if you switched back and forth a lot (which I did for demonstration purposes), you'd constantly see remnants of the modifier. But it doesn't matter because only the first on is used.
![เปิด อาณาจักร หมื่นล้าน "สาระตั้ม" l [Nickynachat]](http://i.ytimg.com/vi/3XE0nPF4YkQ/mqdefault.jpg)
As a newbie to MJ, I find your in depth explanations help my brain grasp the concept. keep up the good work.
Thank you so much for the kind words. You have no idea how happy this makes me. I think my objective here on TH-cam is different from other channels. I want people to come back because they actually learn how something works...in depth. Not just a superficial introduction to some features. I still have a long way to go in terms of production quality of my videos. But I'm trying to improve day by day.
Best youtuber out there on Midjourney. I only got more confused the more I watched others. Christian makes this now much more approachable. I'm trying to find a way to upload architectural models and have them render landscaping and scene without altering my model too much. Haven't figured that out yet, but I see people upload portraits of themselves and generate what still looks like them, so I figure I can do that with a house model.
Wow, thank you so much :)
So right.
@@TokenizedAI really your vids are best about midjorney AI
You're amazing. Keep up the great work. I get excited every time I learn something new that I can integrate into my workflow.
I appreciate the effort, and I find this very useful. I've been using MJ for a while now, and although there are numerous text based tutorials, my mind has never learned efficiently in this way. Having someone instruct me slowly, with consistent examples along the way is the exact way I learn everything (unfortunately). Thank You.
You're very welcome!
Deine Videos sorgen dafür, dass ich immer wieder zurückkomme! Sie sind sehr professionell und informativ. Außerdem ist dein Englisch kaum von einem Muttersprachler zu unterscheiden. Ich freue mich auf zukünftige Inhalte!
Danke! :) Was das Englisch angeht...ich bin größtenteils in Kanada aufgewachsen (9 Jahre). Daher rührt das 😉
You and another guy who calls you out are the BEST WITH MIDJOURNEY. Thanks so much.
Appreciate the awesome feedback so much ☺️
Hi Christian, I just watched your video about Seed and Sameseed, as a newbie in Mid journey, your content is Gold. Thank you so much, you've just earned yourself a new subscriber. ❤️
Thank you :)
you're an excellent teacher, sir. informative, to the point, and so pleasant to hear!
Thank you! 😃
Thank you for your great, calm and in-depth tutorials. I also really appreciate the lack of background music during the tutorial itself. For me, background music almost always interferes with my ability to focus on what is being taught. Of course, this is personal preference, but I appreciate it nonetheless.
My first few videos had music. I think because we're conditioned to think that videos must have music. But I guess it largely depends on what sort of videos you're making.
Thank you so much for explaining the negative prompting! 💗
Loving the videos, You've improved the images I generate by a massive amount! One of the clearest and easiest to follow video series I've found on AI image generation
Thank you so much 🙂
You are a brilliant teacher. You’ve helped massively with how I’m able to create stuff much closer to how I imagine it. Thank you so much. Liked subscribed saved and shared your stuff 👊👊👊
So nice of you to say that 🤗
So well done. I watched both the negative weighting and the God Mode videos. So clear - Thank you!
Glad it was helpful!
Oh man I am happy to thank you for the tremendous effort, I do not know how to return the favor, subscription and admiration is the least thing that can be done for you.
Just keep on learning! 🙂
Your videos on Midjourney are great, thank you for posting them. I find that to be very valuable. You did an amazing job!
Thank you
Thank you so much for these videos, love how thorough they are and easy to follow at the same time. Keep it up! :)
Thanks so much for confirming my assumption that a lot of people would prefer more detailed and nuanced explanations. Have had some rude comments recently with people complaining that they've been too long or that I talk too slow. LOL
Thanks to your tips and tricks, I've been able to save so much time on my daily tasks
Great to hear!
love this series. humble suggestion: pls use ctrl+c and ctrl+v to cut down on typing time. also you can say parameters the way you did before.
Yeah, I've improved that a bit in my more recent videos.
greetings from germany - great work! AND even with my likely basic english I do understand the main part and what you explain. So I would like to thank you very much to teach newbies like me. So funny that you explained that you pronounce(?) "parameter" in a special way. I do copy your style of "parameter" now ;-) because you are a native speaker
Grüße zurück aus HH ;) Ich bin "offiziell" kein Native Speaker. Allerdings bin ich in Kanada aufgewachsen, weswegen ich mehr oder weniger trotzdem dazugehöre ;)
Excellent! This is what I was looking for.
Great to hear!
"Parameter 👍😀"
Thank you for your work!
A lot of useful info.
Appreciate it!
This was an interesting video. I have to say I don’t really have much success using “no” as an editing command. It’s much more for the original composition
Just like with regular prompts, you need to detach yourself from how YOU would describe something. Instead, you need to think about how someone else or many other people might describe what you are trying to do or trying to remove. Also experiment with different words for the same thing. Most people tend to limit themselves by assuming that the AI knows exactly what they mean. When in fact, our description just might not be explicit enough.
My compliments on your excellent videos. One question, I work in black and white, and sometimes I am tormented by mid-journey's addition of grayscale. Is there any way to minimize grayscale?
Great tutorial as always
Thanks again!
your spelling of paramater was great ... for me as an european it sounds much more familiar the first way you've said it ;)
Haha 😂 Well, it appears it's still incorrect though. Since 25% of my viewers are from the US/UK, I might as well bend to their will 😁
Thank you a lot!
You need to redo this tutorial. Up til now I've really enjoyed your tutorials.
Love your videos!
Thank you!
Amazing video!
Glad you think so!
❤ Thank you!!! You are The Best 🎉
Thanks for the content good sir!
You're welcome! 🙂
I WISH that Midjourney would introduce an pinpointing feature!
Care to elaborate what exactly you mean?
@@TokenizedAI Looks like it's my computer editing my words again. What I meant was I wish there was a feature whereby you could edit specific areas of an image, reliably, without affecting the rest of the image.
Hallo Christian, wie stehst du zum Thema Midjourney auf Amazon Merch on Demand? Ich nutze es bereits aber viele haben Bedenken. Ich sehe eigentlich keine Probleme unter Beachtung der Copyrights, welche ich bei Elementen von Vexels auch beachte. (keine VW Bully etc.)
Was genau sollen die Bedenken sein? Die Lizenz zwsichen dir und Midjourney ist eindeutig. Ich wüsste auch nicht, wer konkret Rechte beanspruchen sollte, zumal kein Beweis erbracht werden kann. Wovon ich Abstand nehmen würde, ist die Verwendung der Namen konkreter Künstler. Da bewegt man sich tatsächlich auf dünnem Eis, sofern der Stil ein bekanntes Markenzeichen des Künstlers ist und dies auch ohne AI zu einem Rechtsanspruch führen würde.
Thank you very much... Carry on and good luck 🌺🌺
Thanks! 🙂
What does weight 0 and -1 and below do? do they work? ... what if I use weight 2 or 3 on all elements? is it the same with every number as all elements are weighted the same?
Using a weight of 1 on all segments or a weight of 2 on all segments is the same thing. They are relative, not absolute.
Thanks for you effort, really appreciate it.
Pleasure. And thanks for becoming a subscriber! 😊
Does it matter if you do the varitiation / remix on a thumbnail BEFORE upscaling it? Will that effect the final quality if you variated an image and then upscaled it VS if you upscaled a thumbnail, and then remix that upscale with the same text prompt in both cases?
I don't think can remix an upscale anyway. To my knowledge it only works with regular variations.
your pronunciation is fine, tomato tomahto.
I agree, but hey, I can avoid all of the off-topic comments that way, then I'm fine with that 😅
Many thanks!!
You're welcome!
You said Parameter right the first time. That is how I say it too.
Well, it seems that the majority of the internet disagress with us. I'm happy to olbige either way. Whatever floats people's boat 😅
Fun fact: his voice and videos sound really good still on 1.25x speed if you’re like me and like to really run through all of these
Maybe cause I speak slower than your average TH-camr? 😂
@@TokenizedAI I’m glad you do 😂 I switch back and fourth depending on my attention span
Is there a way to make Midjourney consider a 2 word, word as 1 word?
I am working on a project where the main character have blue hair and golden eyes, and I normally use a reference image.
Where the blue color of the hair tens to bleed into the eyes. If I use the "--no blue eyes". So do it look like it interprets is as "no blue and eyes" making the image a lot more red and still getting the eye color wrong.
Also specifying golden, amber or yellow eyes, seems to have no effect. Since seems to consider it as "golden and eyes"
Color assignment is still a big issue in MJ. I also recommend doing negative prompts via Multiprompts instead, because they are more flexible.
Jesus, I love you're videos! you are a god send! Excellent work!
Wow, thank you! 🙂
Hi, this is Midjourney 4 it seems, is the process the same in v5 and v5a?
Negative prompting hasn't changed.
@@TokenizedAI Thank you very much 😊
Can you do this with the background of an image? Like “no background”
Not really. Like, it won't be transparent because MJ doesn't support alpha channels.
@@TokenizedAI no worries I figured it out. Your vids are helping thanks 👍
it feels like remixing at this stage is so.. useless? should I look at dalle or something else so that alterations would be more to the point?
I tend to agree. If you want to change something very specific, you're better off using Inpainting in DALL-E to make the adjustments. And then use an AI upscaler to get a bigger version of it.
Thanks for the tutorials! I found that --no can sometimes warp the entire output. I was trying to get a boy and his grandfather in cartoon style. It usually generated the grandfather bald and bearded, which is not what I wanted. However using --no beard and --no bald pretty much limited me to always the same exact grandfather and style regardless of changing the style, other keywords or seed. It almost feels like Midjourney considers those features an essential part of being a grandfather? Perhaps using a mild negative weight would work better? I tried different positive words (e.g. clean-shaven) but that had little effect. I ended up completely rewriting and specifying an old man with specific features and dropping the "grandfather" keyword entirely.
Using --no is tricky, for sure! However, negative weights are all the same. Makes no difference whether you use -1 or -0.5. You're asking it to remove something, not just remove it a little bit.
@@TokenizedAI Hmm, it almost sounds like --no is best used when the thing you want removed is almost orthogonal to the main prompt. If it's at all related to the main concept MJ will just tie itself in knots.
hi! many times when i create something it comes with some text or similar to text, many times when i want a logo or a poster midjourney adds some text (or looks like text) HOW TO SAY to midjourney no to add any text? thanks!!!!
Unfortunately it's currently almost impossible to prevent Midjourney from adding text elements. Some have tried "--no text" or "--typography" but that doesn't work because the AI model probably doesn't even know that it's adding "text". It can't remove what it doesn't know. The only real solution right now is to edit the image in Adobe Photoshop or another application and remove the text by hand.
I haven't found that using negative prompts (or even weights) consistently work, especially in regards to signatures and watermarks. Even clicking on v1, v2, etc. for whichever one you want variations on and then editing your prompt to either add the negative prompts or using text weights before resubmitting doesn't always work either. And my prompts are never really complicated because I'm just not that smart. (People shouldn't be correcting you on your English since it's not your native language and most American words come from other countries anyway so in your native tongue you may be pronouncing "parameter" correctly...I swear people need to have better things to do rather than be grammar/language police since at the end of the day it really doesn't matter...I love your accent).
You're right, they're rather unreliable. However, I've found it to be helpful if you use various synonyms for whatever you are trying to remove. Sometimes even words that aren't directly related. So for example instead of "--no signatures, watermarks" you could also try adding "artefacts" to the list and many more. The more I exclude, the better the results tend to be for me.
🔥🔥🔥
Nothing else needs to be said. 😏
🙏🏻🙏🏻🙏🏻
Hi can you give some advice for how to prevent --ar 2:3 streching a figure's proportion? They all look weridly
elongated, or at least a tall supermodel. midjourney will do this for 2:3 all the time, I've tried --no streched proportion --no elongated , but it won't work.
I don't think negative prompts will do you much good in this case. Are you specifically trying to create a character? And are you using Midjourney v4?
@@TokenizedAI yes, it's v4 on default. It's a very common problem, all the other aspect ratio won't have such issue, but 2:3 will produce many elongated characters.
thanks
You're welcome!
Unrelated, but I'm having such a tough time getting mj to create my character. It's a creature that has a human body and octopus head (kind of like Davey Jones in Pirates of the Caribbean). Can you help?
Sounds like a major headache because MJ tends to blend more than we want to. However, you're actually trying to do just that and also have a very specific idea too.
@@TokenizedAI you’re right. MJ tends to either ignore the octopus head instruction or give me an amphibious texture to the body. Do you have any suggestions?
ill take the output quality of MJ over inpainting every day of the week.
Unless you're trying to create something with realistic hands 😂
@@TokenizedAI lol true :) it will come... i hope!
Is there a way to use a negative prompt where you multiple words? As in “--no small dog” to my understanding Midjourney will just take that as no “small” and also no “dog” as two separate negative weights rather than a complete singular prompt
It understands "--no small dog". You can also use multiprompts do it if that feels more comfortable to you.
@@TokenizedAI thanks for the reply, would this work where you are expressing multiple negative prompts I.E -"-no clown nose glasses" in say a situation you have a character you've generated that ends up having a clown nose and glasses, that you want to eliminate, would it not likely take the nose off the character? rather than eliminate the type of nose. I hope that makes sense. From some past previous experience using multiple words in the negative prompt that might describe a singular thing, it sometimes breaks them into separate ideas.
I think so, but why not just do it with a miltiprompt segment. Then it's 100% guaranteed to be interpreted as a negative prompt (not saying it will give you your desired outcome, but the input would at least be correct).
"your primary prompt::1 your negative prompt::-0.5"
@@TokenizedAI thanks for the tips!
Can you make a video comparing mid journey to dalle 2 or is this a mid journey channel
I'm currently focusing mainly on Midjourney but I definitely want to cover DALL-E as well at some point. I just don't find to be particularly good.
I have tried using negative prompting for words/text but it does not seem to work. I am creating an album cover and it insists on adding text to the cover.. Any ideas how to prevent this?
Midjourney is an image AI model, not a typography AI model. It's not capable of generating sensibel text. Whenever you see something in an image that resembles text, then it's only because the images Midjourney was trained on, often contain text. The AI doesn't even know that it is creating text-like content.
How do you know the syntax specifically. So negative is - Why is -- not a positive or is it dash - . If that makes sense. Dash dash -- or colon colon :: isn't intuitive to me
It's not math. It's software engineering. They are very common concepts in that industry, because this how many command-line terminals work.
I know the syntax because there's an official documentation that explains all this in detail.
Para-metre just sounds like a British pronunciation. Silly Americans :) Loving these vids.
How would you write “picture frames”? ::0.5
I'm not sure I understand what your question is.
You can say picture. Or frame. But how would u say a -no picture frame. Does it recognize spaces with a -no prompt
I'm pretty sure --no also recognizes spaces. It basically takes everything until it sees another parameter. Alternatively, just use negative weights.
picture frame::-0.5
@@TokenizedAI awesome. Thanks. I’ll be playing with this very soon. Looks fun to use!
@@TokenizedAI I’ll be testing if the neg weight will include the preceding word. Picture frame::-0.5
Mine came with candles 😢. How do I fix it?
Re-roll?
why is it -0.5 instead of just 0?
You'll have to ask the MJ team
How do you get that remix prompt??
Enter /settings and activate Remix Mode. Then, every time you hit any of the V1-4 buttons, you'll get the remix window.
@@TokenizedAI you rock thank you so much 🙏
Para-meter isn’t the wrong way to say it. it’s just the British way. I’m sorry if someone corrected you, they were wrong.
🔥🔥🔥
💥💥💥
Say parameter however you like. Screw the people that care how you pronounce it.
Hahaha, yeah, though I've not gotten used to it 😅
Absolutely! Say it the way you are comfortable pronouncing it.
great info but please , level your audio please, I had to jump on the volument to turn is ( so ) down during your extra loud intro , then I have had to pull the volume back up to hear back your voice.
I find Remix so unreliable, that it’s not worth using. Even making small changes to a Remixed prompt often causes the resulting images to be very weird (like your birthday cake example). I almost always just input the prompt again, adding my changes. CMD Z (if you just want to bring up the last prompt used) or Copy Paste if it’s an older prompt.
I agree that it's a tricky one. I'm not sure I'd say it's unreliable. I think the problem is rather that we largely still don't fully understand what it does under the hood. From my experience, Remix will stick extremely close to the composition and colors of the image you're remixing, irrespective of the prompt you enter. My guess is that it uses the image as an image prompt and assigns a very high image weight to the prompt, effectively rendering your text prompt adjustments less relevant. I'll have to experiment a little more until I fully understand how it works.
@@TokenizedAI When remixing do you always keep previous prompt in pop-up window? Do you know if there is any impact on the result when you either keep the prompt or just remove previous prompt and type in new words/commands?
So, don't take my word for it, but I believe that when you Remix, it basically uses the image as an image prompt with a very high weight and then uses the text prompt to create something that fits exactly into the composition, shape and colors of the image.
That's why it's pretty unreliable. I recommend trying both. Do extreme stuff to figure out what it's doing.
@@TokenizedAI you have previously replied to my observation.
Should have tried to remove sweetness
Not a bad idea :D
you go ahead and say parameter however and whatever way you want man!
Hahaha. I must admit, I checked and there does seem to be overwhelming consensus that my pronounciation was wrong. The people have spoken :D
When making videos keep in mind the people watch it from a cell phone can't see nothing in here
My more recent videos are better.
Simple comment
Simple response 😆
seriously, the volume in your intro is SUPER LOUD
Yes, this is an old video. Been mentioned many times in the comments. All my recent videos no longer have these issues.
@@TokenizedAI Great! As I appreciate your videos alot. But not being scared out of my computer chair haha. cheers
@@DennisHellberg Funny thing is, it doesn't do that on my system. Yes, it gets louder, but not to the point of being obscenely loud. I think it's just an issue of bad (or inexperienced) sound editing and the fact that everyone has completely different sound setups and default volume settings.
@@TokenizedAI Could be. But then again in the end it's all good. We all learn something new every day, or so they say =) =)
*American Bison
:P :P :P
--no loud intro music
Old video. Was just starting out. New videos don't have this anymore.
Hi, thanks for the content. Just a suggestion, you should remove the mouth sounds between words from your videos. This will improve the production value and be more pleasing to listen to.
Maybe I'm lost in translation but can you perhaps clarify what you mean by "mouth sounds"? I think I know what you mean, but I'm not 100% sure 😅
lol!!!
not helpful -
Yes. Your message was received, lol 😂
too much talking get to the goods bro
too much impatience and entitlement so figure it out for yourself bro
(the state of attention spans in 2023)
i wasted my free trial
😞
Not your best tutorial
Can't win 'em all. 😉
hey guy - come to the point - just a hint for your vids - make it short - and focus on the information to want transfers - hot helpful I would say - so thumbs down
I appreciate the feedback. Best of luck finding what you're looking for.
Muy bien explicado gracias
wait, 4 weeks ago you where still using --v 3? Okay seriously, visit prompt-chat WAAAAY more
lol, no I wasn't. I was merely illustrating the difference between 3 and 4. Should be obvious enough from the video.
3 weeks ago v3 wasn't set as the default yet. So if you switched back and forth a lot (which I did for demonstration purposes), you'd constantly see remnants of the modifier. But it doesn't matter because only the first on is used.