Stable Diffusion - Mastering Camera Angles & Shots in AI Image Generation
Вставка
- Опубліковано 8 сер 2024
- Do you struggle to get the right shot? In this video I go over framing your shots and getting the right distance. How aspect ratios effect those images. These tips are useful for all AI image generation tools including Stable Diffusion, Midjourney, Dall-E and more. How to get that full body shot. #Fooocus #StableDiffusion #DigitalArt #ImageManipulation #CreativeTools #midjourney #Dall-e
Some of the ones I have compiled: ai.kleebz.com/shot-types-and-...
I am running on an i5 32GB RAM Nvidia 3070 8GB VRAM
If you find the video useful and would like to tip, UA-cam does have the Super Thanks option. Those are greatly appreciated so I can have my coffee!!
☕ BUY ME A COFFEE: ko-fi.com/kleebztech
Also I am a verified creator with Brave and do accept BAT for users of the Brave browser.
00:00 Intro
00:47 Aspect Ratios
06:42 Distances
09:00 Angles
10:10 Other shots - Наука та технологія
A list of some of the ones I have compiled: ai.kleebz.com/shot-types-and-camera-angles-for-ai-image-generation/ If you find the video useful and would like to tip you can buy me some electricity for all these image generations. It is greatly appreciated! ko-fi.com/kleebztech
Impressive the knowledge of Fooocus you display in your videos, you are certainly the go-to channel to get information about it.
Thank you!
Man, i was looking for this video. You don't know how much you helped me. Thank you very much.
Very welcome!
Pausing halfway to say wow - this video is gold. I'm subbing for more informative, useful, concise vids like this.
Awesome, thank you!
Understanding how camera angles, distances, and aspect ratios interact is crucial for achieving the desired results in image generation.
Merci pour ces précieux conseils qui nous font avancer ?
After watching your vids, I realized I can do almost everything I need in fooocus and photoshop. Even a1111 seems like overkill.
For most things you can. I still use the other ones but not very often. I spend too much time tweaking things in those it seems.
That's a great lesson! Thank you!
Thanks for leading the way.
Thank you, keep it up. I have wildcard text file with shots. After this video I added a few more. Thx again.
I never thought of that. That could be interesting.
AMAZING tutorial. Thanks a Lot!
You're welcome!
Hey bud we appreciate lessons, keep it up!!👌
Thanks! Will do!
Good work.
Thanks!
This could me great in styles😮
Hey Rodney, I've been watching most of your videos on fooocus and loving them so far. I would love to see some content focused on inanimate objects, such as product images? It could add an intriguing dimension to your work.
Looking forward to seeing what you create next!
Are you talking about trying to incorporate actual product images into AI generated images?
@@KLEEBZTECH Ye, so lets say I have a photo of an iPhone and I want to incorporate that into an ai generated human holding it or doing some action with it.
Problem there is that Fooocus itself would probably not be able to do that right now very well. But I will experiment. I do know there is a new AI project that is actually concentrating on that. Can't remember the details.
awesome stuff
Thank you. There is much more when exploring this topic. Check out photography websites for terminology. It can really help.
remarkably interesting study, thanks a lot. If you can make some videos about the different lightning and face expression, it will be perfect. I think is that make the bester pictures.
great video!
Thanks!
Thanks teacher
You are welcome
Thank ❤
Welcome!
Thanks. I think a video that converts photos into anime and vice versa is interesting
Anime into a real photo.... Interesting. Not sure how well that would come out but I might experiment with that.
@@KLEEBZTECH I found some "workflow". Use the juggernaught model and go load the anime Image into the Inpainting and the Describe Tab. Click "Describe Anime" to get the prompt. In the Inpaint tab, mask the complete picture and use the "Improve Detail" while leaving the "Aditional Prompt" empty. This should do the trick :-)
Interesting. I'll add that to my list of things to try out. 👍
Started playing with this. Interesting. Now I have some more ideas to explore with this... 👍
And here it is. Thanks for the idea! ua-cam.com/video/PuC2AVrQXTs/v-deo.html
Great Tutoriual, Helped me a LOT!!!!!!!!!!!!!!!
Very welcome!
👍👍👍👍
Amazing .. do you have more "promts" tutorials?
Not yet.
First of all let me express my huge gratitude for these tutorials and especially for this priceless topic!
Now, I also have a question. In your opinion if the current AI imaging like Fooocus or stable diffusion requires at least 6 gb of VRAM (Nvidia case), would it be wiser to get a pc with 8 gb vram video card now, like nvidia 3070, or to purchase a 16gb or 24 gb vram pc? Im asking because Im wondering which road to take and how future-proof it is and whether its worth it. Im also pondering when will text to video AIs appear and what would be the better decision - to get an expensive pc now, a somewhat expensive pc, or an acceptable pc, in terms of components? What would you do and whats your opinion if you had to purchase a PC now for AI imaging and eventually for text to video? Im asking because my hardware is really lacking and Im wondering which way to go and how much to invest (whether its even worth it)
If you were getting something now I would not settle with 8GB VRAM. The trend is needing more VRAM and can imagine video will be much more intensive. I actually have doubts that video will be easy to do on your own system. Even with my 3070 8GB I have considered upgrading but right now am waiting to see how things play out before I do.
I can recommend a 4060 16GB. It is "cheap" and has lots of VRAM. All other components did not make a huge difference.
Someone needs to build a spreadsheet of angles, lenses, lighting etc! There is just too much to remember
I put up a list here of what I have compiled so far but there are many more: ai.kleebz.com/shot-types-and-camera-angles-for-ai-image-generation/
Aks ChatGPT for these lists ;-)
If you get lucky and it isn't lazy and only gives you a very short list. Lol. Damn lazy AIs. I've used them also for wild card list but even there they usually won't give me as many as I want.
@@KLEEBZTECH i alway write something like "give me 30 answers"
@@metanulskibe careful you might break it. 😊 Sometimes I think it depends on what day or time I ask it
What is the aspect ratio equivalence of 16*9 in fooocus? 7*4?
Yup.
Wanted to ask about the upscale. I tried upscaling in fooocus to 1.5x but it gave me a new character. If I use the same seed, would this resolve the issue?
The first two upscale options also vary the image. Fast upscale will not but also will not add much detail. You can use FaceSwap with the vary and upscale as well if you enable that in debug mode. I think you can also change the "Forced Overwrite of Denoising Strength of "Upscale"" to minimize the changes but I have not played with that much yet.
Thanks for the tips. Unfortunately Dutch Angle Shot doesn’t work on realistic photography models in SD
You can also try something like canted angle shot.
Which model you are using in fooocus
Usually JuggernautV8 including this video.
wats the speed of rendering 1024x1024 image with your 3070 card?
Quality settings (60 steps) about 35-40 seconds total to generate and save.
Hi new here I usually create pictures of people what I want to do is to be able to fit a person into the correct aspect ratio but have them from head to toe to fit exactly into the frame I find it hard because I’ll get half a body or a head it never uses the aspect properly how I want it so a person standing from head to toe in the frame
That can be difficult to get exactly into frame. But if you get images that have the whole body with some extra around edges you could always upscale and then crop to the aspect ratio you want I suppose with image editing software.
I’ve tried putting words like 1 foot away or 2 foot away or 3 foot away but I don’t think it understands these commands
No it won't. Not sure what you are using but if SDXL you could try this LORA. I have not tried it so not sure how well it works. civitai.com/models/213307/envy-zoom-slider-xl-01
8:28 prompt weight
Do you have a question about prompt weight?
@@KLEEBZTECH no, it's just a note for myself, so I can reference that part of the video later. Thanks
Is it possible to make realistic AI influencer by using only fooocus!?
I know someone who does.
Yes
@@KLEEBZTECH I created one!!
Now I know two that have! 😀
@@KLEEBZTECH 😁 👍
👍👍👍👍