Welcome!
Get comfy, because now you're going to be the director of your own movie~.
1. The box where magic happens
Metropolis
(writing)
Models Used on Metropolis
CinemaScope
While i was happy with the capabilities of v1.0, still want to improve aspects such as backgrounds and lighting, so i was testing new kind of merges and i think this is veeeery nice, i've been learning a lot from other creators these last months and its been really fun.
This new version, brings a lot of new stuff about backgrounds, lighting, scenery and most important, prompt adherence, so give it a try!
If you want interesting gens, try widescreen proportions!
@zippyman2929 is a very skilled prompter, really good one, so in the need of giving the highest quiality showcase i ask him for help. This showcase is a mix between our works, so give him a lot of kudos and hugs!.
Models Used on CinemaScope
l'usine Lumière
After some time learning the magic of merging, i can say finally i have a personal merge that suits the path that i'm looking for.
Presenting to you, Cinema is a model that focus on non realistic furry characters over semi-photorealistic ambients with heavy handling of lighting, all of this for pursuiting a more "Cinematic" aesthetical feeling.
Of course, one of the thing i always pursuit on checkpoints are prompt adherence. This one is very very good imo, works amazing with e621 tags which are the ones i'm used to using.
This is a first version for testing, i like it how it came and hope update through time.
Models Used on l'usine Lumière
2. Recomended Settings
Steps: 25–40
CFG Scale: 4–7
Sampler: Euler a (Can be used with DPM++ 2M Karras, but haven't tested enough)
VAE: Integrated
Prompts: Use e621 tags!
Upscaling: I recommend 4x_NMKD-Siax_200k, But others can be used too
Denoise: 0.2-0.4
Steps: 10-25
CFG: 1
Positive Prompt
For high-quality generations:
masterpiece, amazing quality, best quality, detailed background, newest, scenery, highres, absurdres, furry, anthro,masterpiece, amazing quality, best quality, detailed background, newest, scenery, very aesthetic, ultra-detailed, 4k, volumetric lighting, ray tracing, realistic, photorealism, highres, absurdres, furry, anthro,For general use with LoRAs
masterpiece, amazing quality, best quality,Negative Prompt
worst quality, old, early, low quality, lowres, signature, username, logo, bad hands, mutated hands, glowing eyes, simple background,Lighting
blue lighting, dark lighting, red lighting, purple lighting, yellow lighting, and you get the idea...Hope you like it people! Have fun! ❤❤❤
Description
FAQ
Comments (19)
Hello? Hello? Yes, is this the President of The Republic of Peak?
This cinema model thingy looks AMAZING, the images are so CLEAN (I'm jealous) I am absolutely stoked for this.
Kskdjaks Thaaanks
Hope you having a good time enjoying it UvU ❤️❤️❤️
This model really lives up to its name, it's the first AI model on here in a while that's legitimately impressed me with how great everything looks! Well done! ^v^
absolute cinema
Sup people, Lumiere (v1.0) will be on-site this week, next week and onwards i will focus on CinemaScope mainly,
Thanks for all your support people ❤️❤️❤️
This model is blessing...
And this model is the world!
Worship it! Give praise unto it!
It's beautiful and sublime!
It's impeccable, almighty and divine!
All hail Cinema!
Thaankiee ❤❤❤
This is just a another merge with a purpouse behind, with a lot of work from others creators before, just another step in this fun and exciting road of ILL hahaha, kudos for all that people who contributed to reach what we have today ❤
Hold up is this glaze justified?!
@TheGarandMann I mean its a merge, and the component of this merge are others merges, see if it were a dna, it has a little piece of every other merge that came before :3
@SXCD2719 Yeah fair I suppose, that's what I call the Habsburg effect, but it's actually peak instead of birth defects :)
@TheGarandMann I just was making a Dragon Ball reference. :3 (Zamasu's quotes are always over dramatic)
EDIT: Only for CinemaScope
HEY EVERYONE
Are you having trouble with 4 finger gens instead of the wanted 5 fingers? just put "fewer digits" on negative prompt JDKAJAKDJKDJKSD
Updated description
Thanks!!!
it worked for me x3
@SXCD2719 Thanks for helping with prompting :)) Maybe you should do a guide or something for pos. and neg. prompting
Yo random question, how do you all assemble your prompts? One more plea to those who are much more skilled than me, just before the year ends. I'm also so sorry if this seems long.
for me, it's a little something like this:
No. 1, Image design tags: "Masterpiece, best quality, amazing quality, detailed background" Just to name a few.
Line break is ALWAYS mandatory in my prompts (though I'm sure it's not needed)
No. 2: Person, pose und clothes (epic rhyme)
"1girl\boy, solo, [Species Here}, [Fur Colour Here:1.2), [Eye Colour here], [Complimentary, but not needed stuff, pawpads, fangs, etc]. Then, emotions [smile or something] continuing with actions ("looking at viewer" and "dynamic pose are always in mine"). Lastly, [Clothing here] button shirt, jeans, etc..."
Epic line break
No. 3: body shape
[Chubby, slim, thick thighs, whatever really, maybe some words are numbered like the fur colour if I want extremities]
One last line break!
No. 4: the general area
[Cafe, park, beach, maybe an addendum like cyberpunk if I want a specific style]
I just sorta feel like I'm doing something wrong, nevertheless, I hope I can get some help from one of you guys. :)
Oh and Merry Christmas as well!
If you wanna talk about this deeper, send me a dm and i can teach you~
@SXCD2719 That sounds quite...sensual, but sure
Depending on the checkpoint, I add "furry, anthro" to the quality tags, as well as style tags if the model works with them.
Then it's a short description of the scene, if more than one is involved (3 naked cows in the moonlight), followed by (1girl, male/female furry, species + sex, body type, fur colour, hair colour, hairstyle, eye colour(s), clothing + extras), description of the vile depraved yet fun act in question, (1boy/whatever, same as before but for the other other one), followed by camera and image instructions (xyz focus, closeup, multiple views, crosssection, motion lines, you have it).
Take note that the two hot cotenders for this year, Qwen 2512 and ZImage, use "natural" prompting, which while it gets shot done far more precisely, requires me to write a 2-page elegie on the nature of things as depicted, and doesn't come with a decently working negative window. Both models can't do furry or yiffy stuff worth shit at the moment (my attempt at a cute anthro goat via Qwen will haunt my dreams for a while), but I have high hopes that someone trains Zimage on the danbooru/e621 stuff for the ultimate furry checkpoint.
Until then Illustrious it is. Yiff for the Yiff gods, Smut for the Smut throne and all that.
P.S.: same offer as the other guy, if you want to wax poetic about prompts, feel free the hit me up. I don't bite.
Dude, it's not that simple. Depending on the model and language, things can vary. Even the same models on Illustrious can interpret the same hint differently.
Let's say in one model you can create a scene with several different characters of different colors and get more or less the right result, but for another you'll have to find a different approach.
Also, don't forget that for a great picture you'll have to make corrections (Photoshop, ADetailer, img2img) unless it's something abstract.
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.