I didn't know how to find the seed of already generated images, great tip. Also, I'm glad to see you have a sponsor! I've been trying v5, but so far I've been busy enough just comparing it to v4, regardless of parameters. But it's good to know which parameters are working (or not) at this moment, to avoid losing time running tests for nothing. Thank you so much for another great video! 👏👏👏
@@MakingPhoto yeah and v5 is getting better and better, after a dodgy start (for me) where I wasn't overly happy with the output I am now extremely happy with it
🎯 Key Takeaways for quick navigation: 00:00 🚀 Mid-journey version 5 is out for Alpha Testing, featuring realistic images. 02:48 🎨 Dash dash tile parameter creates repeating patterns for backgrounds and designs. 03:28 🖼️ Aspect ratios significantly upgraded in version 5, providing a wider range of options. 04:24 🎨 Stylize parameter lets you control Mid-journey's unique style from 0 to 1000. 05:19 🔀 Dash dash style parameter differentiates between style variations. 05:46 🔀 Dash dash chaos parameter controls diversity in grid images. 06:14 🚫 Dash dash no parameter instructs Mid-journey on elements to avoid. 06:42 🔢 Image weights are back, allowing emphasis on included image URLs. 07:25 🎥 Dash dash video parameter not functional in version 5 for process videos. 07:53 📊 Dash dash Q parameter for image quality is disabled in version 5. 08:19 🌱 Dash dash seed parameter defines a starting point for image generation. 09:08 📐 Upscaling parameters not active in version 5 yet, await updates. 09:49 🔄 Mid-journey version 5 is in beta, some parameters might improve over time. Made with HARPA AI
V4 merges more the imagination with the cinematic touch. The realism in v5 is nice but its more photography with a bit of surrealism. I think commercially better imo.
I think you're right. I've been bouncing back and forth on every prompt just to see how v5 handles things differently. I'm not yet settled on which I like best.
I find your video very helpful, thanks a million, I use image prompt mixing with text prompt, I find MJ hard to pick up my ref image face and to use in the output. Is it really possible, or MJ just void it for ethical issues?
Thanks! If you want an exact face swap, Midjourney won't do it. You'll have to use Photoshop. Midjourney is designed to blend - always blend and mix. So, the face changes. I have a video on making avatars with a few tricks to keep the likeness if not the exact face.
If you're talking about simple feedback to Midjourney, yes! Use the rating system (it's in flux and either a heart or emoji faces). If you mean to change the image, then use the Remix tool. Type /Prefer Remix in the message bar. Enter. This turns it on. Now, when you make variations, you can change details. But there's a limit to how much you can direct the AI.
Where is the best place to go to get prompts? As a beginner who is trying to create graphics for my Etsy digital art initially in MidJourney it seemed fairly easy to get good outcomes, now it is a nightmare.
You can still get great images in MJ. Have you tried using a prompt helper? I have a video covering some of my favorites. th-cam.com/video/Pwtn9AKOBb8/w-d-xo.html
If I don't give you a prompt, it's usually because it was an accident (an image I like, but not what I was expecting from the prompt). Here's the prompt: Robot dog, advanced technology, sleek metal body, glowing blue eyes, futuristic cityscape, bright neon lights, high-tech, advanced, eerie, In the style of H.R. Giger & Zdzisław Beksiński, digital, --chaos 80 --ar 16:9 --stylize 600 --v 5
@@MakingPhoto Nope, I tried forcing it into v4 through settings and the --v 4 flag as well. No dice. Im using a 3rd party face swap tool now, but it isnt free.
@@jwulf Ok. Yes. Photoshopor the like is best for face swap. The AI will ALWAYS try to blend and change your face. It's generative. Usually, when you reroll a few times, you'll get likenesses - though if you want a face swap, AI isn't the tool. Good luck!
Here's what ChatGPT has to say to your comment: Hey there, thanks for your feedback! We understand that vocal fry may not be everyone's preference. We'll definitely consider this as we continue to create and improve our content. We appreciate you taking the time to share your thoughts with us, and we hope you find something enjoyable in our other videos. Have a great day! 😊
--no is a weird parameter. Sometimes I think it's working great. Other times, it seems to do the opposite of what I expect. V5 changed up the parameters. --stylize works much better in v5. I'll keep checking to see about --no
![[Full] 4 ต่อ 4 Celebrity EP.910 | 22 ก.ย. 67 | one31](http://i.ytimg.com/vi/4TR3MVpOiw8/mqdefault.jpg)
Thank you for the video listing all the parameters in your description!
No problem! I was getting confused, so I thought it was time for a recap.
I didn't know how to find the seed of already generated images, great tip. Also, I'm glad to see you have a sponsor!
I've been trying v5, but so far I've been busy enough just comparing it to v4, regardless of parameters. But it's good to know which parameters are working (or not) at this moment, to avoid losing time running tests for nothing.
Thank you so much for another great video! 👏👏👏
Awesome, thank you! They added the seed to the website, but quietly. Otherwise, it was send yourself a message in Discord.
Thanks for your hard work. Great editing, great explanations.
Glad you enjoyed it, Christian! Keep creating!
You deserve a lot more likes. Very well done, short explanations but still clear.
Glad it helped!
love your vids keep them coming thank you
Thanks, Wayne will do!
I'm glad you mentioned stylize as I'm also not sure it's working properly as I've noticed very little difference between s50 and s750
I'll give it a bit and then test the parameters again. Sometimes when a new version comes out, not all the parameters are working.
@@MakingPhoto yeah and v5 is getting better and better, after a dodgy start (for me) where I wasn't overly happy with the output I am now extremely happy with it
Part of this is the program, and part is that I'm getting used to the prompting.
Hi there I enjoyed your video tutorials.
Can you make a video on how to write a prompt on Midjourney to create backdrops for clip arts. Thanks you.
🎯 Key Takeaways for quick navigation:
00:00 🚀 Mid-journey version 5 is out for Alpha Testing, featuring realistic images.
02:48 🎨 Dash dash tile parameter creates repeating patterns for backgrounds and designs.
03:28 🖼️ Aspect ratios significantly upgraded in version 5, providing a wider range of options.
04:24 🎨 Stylize parameter lets you control Mid-journey's unique style from 0 to 1000.
05:19 🔀 Dash dash style parameter differentiates between style variations.
05:46 🔀 Dash dash chaos parameter controls diversity in grid images.
06:14 🚫 Dash dash no parameter instructs Mid-journey on elements to avoid.
06:42 🔢 Image weights are back, allowing emphasis on included image URLs.
07:25 🎥 Dash dash video parameter not functional in version 5 for process videos.
07:53 📊 Dash dash Q parameter for image quality is disabled in version 5.
08:19 🌱 Dash dash seed parameter defines a starting point for image generation.
09:08 📐 Upscaling parameters not active in version 5 yet, await updates.
09:49 🔄 Mid-journey version 5 is in beta, some parameters might improve over time.
Made with HARPA AI
So far I personally like v 4 more. The photo realism is cool, but the artistic style 4 mastered has far more commercial use.
😃😍I have a soft spot for the v4 aesthetic, too.
V4 merges more the imagination with the cinematic touch. The realism in v5 is nice but its more photography with a bit of surrealism. I think commercially better imo.
I think you're right. I've been bouncing back and forth on every prompt just to see how v5 handles things differently. I'm not yet settled on which I like best.
I’ve had some mixed results with v5. The aspect ratio of 16:9 sometimes works, most of the time it doesn’t.
Yes. v5 is still changing. I haven't noticed a problem with 16:9, but some of the other AR aren't quite mathematically correct.
I use 9:16 a lot I'd say about 25% give me a weird result but it's still very useful for me
Hmm. I don't notice a problem with 16:9. A little stretchy. It's probably my style of art.
I find your video very helpful, thanks a million, I use image prompt mixing with text prompt, I find MJ hard to pick up my ref image face and to use in the output. Is it really possible, or MJ just void it for ethical issues?
Thanks! If you want an exact face swap, Midjourney won't do it. You'll have to use Photoshop. Midjourney is designed to blend - always blend and mix. So, the face changes. I have a video on making avatars with a few tricks to keep the likeness if not the exact face.
Is there a way to give it feedback/notes on an image generated in order to further refine the creative direction and details?
If you're talking about simple feedback to Midjourney, yes! Use the rating system (it's in flux and either a heart or emoji faces). If you mean to change the image, then use the Remix tool. Type /Prefer Remix in the message bar. Enter. This turns it on. Now, when you make variations, you can change details. But there's a limit to how much you can direct the AI.
@@MakingPhoto Thanks!
Where is the best place to go to get prompts? As a beginner who is trying to create graphics for my Etsy digital art initially in MidJourney it seemed fairly easy to get good outcomes, now it is a nightmare.
You can still get great images in MJ. Have you tried using a prompt helper? I have a video covering some of my favorites. th-cam.com/video/Pwtn9AKOBb8/w-d-xo.html
what is the prompt for the wolf in the video ?
If I don't give you a prompt, it's usually because it was an accident (an image I like, but not what I was expecting from the prompt). Here's the prompt: Robot dog, advanced technology, sleek metal body, glowing blue eyes, futuristic cityscape, bright neon lights, high-tech, advanced, eerie, In the style of H.R. Giger & Zdzisław Beksiński, digital, --chaos 80 --ar 16:9 --stylize 600 --v 5
No matter what I try, I can not get it to integrate my face with image to image in version 5. Do you know why that could be?
How many images of you are you adding to the prompt?
@@MakingPhoto Started at one, and have gone all the way up to 5.... the person it makes barely looks related to me LOL
@@jwulf Hmm. Does it work in v4?
@@MakingPhoto Nope, I tried forcing it into v4 through settings and the --v 4 flag as well. No dice. Im using a 3rd party face swap tool now, but it isnt free.
@@jwulf Ok. Yes. Photoshopor the like is best for face swap. The AI will ALWAYS try to blend and change your face. It's generative. Usually, when you reroll a few times, you'll get likenesses - though if you want a face swap, AI isn't the tool. Good luck!
a tad less vocal fry would be great
Here's what ChatGPT has to say to your comment: Hey there, thanks for your feedback! We understand that vocal fry may not be everyone's preference. We'll definitely consider this as we continue to create and improve our content. We appreciate you taking the time to share your thoughts with us, and we hope you find something enjoyable in our other videos. Have a great day! 😊
--no is def more potent in 5.2 it's broke
--no is a weird parameter. Sometimes I think it's working great. Other times, it seems to do the opposite of what I expect. V5 changed up the parameters. --stylize works much better in v5. I'll keep checking to see about --no
You just earned a sub from me
Great! Enjoy creating!