Midjourney Panning + Dreamshaper Version 7
Vložit
- čas přidán 3. 07. 2023
- Midjounrey Panning offer a crazy new way to create very very wide or high images. You can create the most impressive Panoramas ever with the new Midjourney 5.2 function.
Dreamshaper Version 7 is a beautiful Model that can do everything from Photorealism to Digital Art and Anime Style. All of that in a very high quality.
#### Links from the Video ####
Dreamshaper 7 civitai.com/models/4384?model...
BadDream + UnrealisticDream civitai.com/models/72437?mode...
Phoenixdress civitai.com/models/48584/phoe...
Detail Tweaker civitai.com/models/58390/deta... - Jak na to + styl
Yo, what an honor!
If you like realism, you should try Absolute Reality too (very similar to dreamshaper, it branched out of there, but more realistic).
Absolute Reality + game of thrones lora is probably the most realistic sd1.5 can get.
I also made a detailer lora myself and a fastnegative embedding mostly to be used with anime and art. Cheers! And thanks for featuring DreamShaper!
Yo! Great to see you here!
Love your work, DS and AR are my top favorite models!
Thanks for the tip with the Lora! Will try out as soon as I'm back home :)
Excellent, actionable information as always. One of the few must-watch aitubers.
Man I have been making tedious versions of this... so glad you are covering this... I have wanted this to be automatic for so long
thank you :)
You really should do a video on Civitai helper extension. Worth the shout out, and speeds up download, model management.
Dreamshaper 7+LeonardoAI+Alchemy I need this now!
Cool shirt! I've had the same shirt for years now 😅
SICK AF HOMIE! 💯
thank you
you video are so dam good! Keep up the good work
Thank you so much
@@OlivioSarikas no problem Olivio.
The only issue I find with negative embedding is that it seems to limit the models promoting accuracy. I have found that adding it last can for some reason fix alot of the issues with that for some reason. It can also make upscalers do some weird things at even low denoising values.
Is there any video tutorial about how to use that?
If I have a LoRa trained with clip skip 1 do I need to retrain for dreamshaper?
Double brackets are the same as weight 1.21 btw, so ((abc)) equals (abc:1.21). Every bracket multiplies the weight by 1.1. I like to have everything in numbered weights, just for the sake of tidyness. :p
can you do a video on SuperMerger please, the reason for that 1 i want use my trained face with diffrent models like cartoon, anime realistic 2 you can't train most models outside of A1111 because most of them need to convert to diffuser to be able to train but when you merge them with trained model than you can use it to train your face
Hi Olivio.. Can you please make a video on local installation of DragGAN please???? 🥹
Each time you use pan you lose resolution of the origin picture am i right?
How do I have to set that in Midjourney so that I get a real panorama picture. So without a step in the picture.
For some reason I cannot load Dreamshaper 7, throws up an error: changing setting sd_model_checkpoint to dreamshaper_7.safetensors: SafetensorError
can u show install stable diffusion using amd gpu i cant do it
What software did you use to implement the long tracking shot across the image (at the beginning of the video)? It's so wounderfull smooth 😀
midjourney. or do you mean the panning from left to right? that was davinci resolve
@@OlivioSarikas ah davinci resolve - Thank you 😁😁🤩🤩
Hello. Could you please give me a recomendation on how to draw mandalas with AI.
Thank you
Different models like different prompts. Is there an addons to automatically load certain prompts to the prompt box when you change model in A1111? It's difficult to remember each models preferences when prompting. Same question but to model settings too like certain models like clip skip 1 while others 2 so how to remember all this information?
you can save them as styles and load them when needed
@@OlivioSarikas amazing. Thanks. Will that also enable me to pre-save things like the clip skip number? I'll look into styles and get familiar with it.
I’m curious if the other options appear again if you choose the “make square” option.
interesting idea :)
They do, but then you go back to a 1024x1024 image, so you lose a lot of resolution.
Any idea how to get the one prompt extension to allow more than 50 generations at a time?
set everything up as normal in the extension, then use the normal generate button and the batch count instead. You also need to set the render settings (size, steps, cfg and such) in the normal settings, not the extension
@@OlivioSarikas ooooh! Wow, I had no idea. Thank you!
is there a way to use 2 LoRA at the same time?
yes, lower the weight of each lora until the result looks good
I sometimes use up to 4 loras even. With the regional prompter extension, you can go even higher as you can assign loras to certain regions in the image only, where they won't interfere with others.
Perhaps arthemyComics, or revAnimated for another model review.
Dreamshaper 7 already out? They just released 6 😮
That model is going to the moon :)
@@OlivioSarikas Indeed! Dreamershaper is great for stylized art but struggled with realism. This one seems to fix that!
@@ATLJB86 the creator also has a realistic model called AbsoluteReality. It's pretty good too :)
Dreamshaper 7 looks like Midjourney 3
Yeah you keep struggling with no control whatsoever, mj is for amateurs.
Stable Diffusion Models are really behind Midjourney. I used Midjourney long time ago after it got two models to pick for first time but already back then it was more controllable in terms of results through prompt. Control net is super and few other plugins but Models iyltself are very poor. You can get OK results but never what you really want but what author of model had on his mind while training his model. Very narrow results.
2:48 That's unrealistic! You would be able to see the camera in the reflection in the armor! Totally won't fool anyone!
/s
Maybe inpainting could help..
really? you have never seen a shot with a hidden camera? not watching movies much?
@@OlivioSarikas that's done with inpainting 😁.
Here's one better , learn to draw