Este modelo es el resultado de un merge experimental entre CyberRealisticPony_v90Alt1 y NevolutionSDXL_v2Base, dos modelos SDXL con enfoques muy distintos pero altamente complementarios.
VAE no incluido, use SDXL_vae
La idea detrás de esta fusión fue combinar la precisión en rostros y figuras humanas de CyberRealisticPony con la potencia creativa y riqueza fantástica de NevolutionSDXL. El objetivo es ofrecer un modelo versátil capaz de generar tanto escenas realistas detalladas como composiciones oníricas, surrealistas o épicas sin sacrificar calidad en ninguno de los dos extremos.
Es mi primer merge, así que toda crítica constructiva y feedback son bienvenidos. Espero que lo prueben, experimenten con él y, sobre todo, que les inspire tanto como a mí al crearlo.
¡Gracias por pasarse y feliz generación!
This model is the result of an experimental merge between CyberRealisticPony_v90Alt1 and NevolutionSDXL_v2Base, two SDXL models with very different but complementary strengths.
The VAE is not baked, use SDXL_vae
The goal was to combine the highly accurate human figures and facial details of CyberRealisticPony with the fantastical creativity and visual richness of NevolutionSDXL. The result is a versatile tool capable of producing both lifelike portraits and dreamlike, surreal, or epic scenes—without compromising on detail.
This is my first custom merge, so I'm open to feedback, impressions, and suggestions. I hope you enjoy using it as much as I enjoyed creating it.
Thanks for checking it out, and happy generating!
Description
Incorporated a bit of IlluSatPony-Exp to refine it. I really dont know how to tag this model anymore, it can be use the scores frome pony promtps, but some pony loras may break the outputs, so im inclined to tag it as SDXL. It doesnt have that much Illustriuous to to tag it as one.
Better results with 40-50 steps.
Se incorporó un poco de IlluSatPony-Exp para refinarlo. Ya no sé cómo etiquetar este modelo; se pueden usar las puntuaciones de los mensajes de poni, pero algunas loras de poni pueden dañar las salidas, así que me inclino a etiquetarlo como SDXL. No tiene tanta Illustriousidad como para etiquetarlo como tal.
Mejores resultado entre 40 y 50 pasos.
FAQ
Comments (13)
trying to learn to use Super Merge extension right now... i see it have a lot of otpions to control the merges more eficiently. Any sugestion (or tutorial link) is welcomed,
As far as learning and improving goes I find ChatGPT to be very helpful there, you gotta give it a lot of info like "Im using webui, Im using supermerger, I'm merging SDXL models, I want to transfer realism from one model to another, can you help me do that through targeted advanced block weight merge, it can tell you what blocks contain what info, help you get more surgical only taking what you want from the donor model and avoiding affecting the parts of the model you dont want to mess with, so Id suggest trying that out
For your case in particular, I would start with the fantasy model as the base and transfer the realism from a realistic model, or multiple realism models over to the base fantasy model, it could work the other way as well, but targeting fantasy elements would be difficult, its much easier to target realsim and bake that into the fantasy model
@Urameshi thanks for the advice. Yeah i, using chatgpt, but sometimes a real person is needed :P
overall i like this, as the faces look pretty real including many background guys. But it lacks creativity and doesn't quite follow prompts. Like I want people in the background but they are often isolated. Fingers are random, sometimes I get that illustrious finger and sometimes I get a handful of fingers. I wonder if you mixed it with juggernaut or something if it would give more creative looks, but keep the faces looking pretty real.
Thanks for your coment im already working on making it better, can you share some of your generations with the promts? it helps me a lot.
I'll try to remember, I was doing a cfg test last night and ran out of time, I found cfg 2 worked best for me, I still get some weird faces. I found illustrious is terrible at steampunk, which is what I mostly want to make, sdxl is better. And I had one weird prompt I took from this place that is supposed make something that looks like a bee on an eye lid, which I always get weird results in. But for this one, there was always a bee and a ladies private parts every time
i think it was this prompt: which was gotten from this site:
Illustrate a surrealist, macro-level close-up of a human eye with a perfectly formed honeycomb pattern fused into the iris, where the hexagonal cells appear to extend inward like a tunnel of light and order, each chamber subtly pulsing with golden illumination. The pupil rests at the center of this hive tunnel, a dark portal surrounded by symmetry and structure, suggesting a balance between organic instinct and intelligent design. The skin surrounding the eye is covered in a biomimetic mesh resembling beehive architecture—hexagonal tessellations that stretch like scales across the lid, brow, and upper cheek, blending seamlessly with natural skin tones. A single honeybee lands delicately on the lower eyelid, its tiny wings catching the ambient light, its fine legs gently touching the boundary between the hive eye and human skin as though delivering a secret. The bee is captured in perfect anatomical realism, its eyes reflecting the honeycomb pattern it seems drawn to. The eyelashes curve like antennae, slightly iridescent at the tips, creating a unified aesthetic between insect and human. Lighting should be dramatic but intimate—highlights glinting off the wet iris surface, shadows forming between the skin’s honeycomb ridges, and a subtle lens flare in the reflection of the bee’s eye. The concept should convey the union of nature and consciousness, where the eye becomes the central organ of the hive—a visual metaphor for collective intelligence, ecological awareness, and the quiet, powerful intelligence of pollinators in a shared biosphere. Let the entire composition radiate reverence for the elegance of bees, the miracle of vision, and the hidden geometries of nature woven into the human form.
And where it gets vagina from I don't know but every time I made it, it came out like that.
you can also try mixing it with epicjuggernautxl_vxvXI that one is very good with quite a few things especially architecture. And eCommercePhotography_v10 for steampunk
@frankmike nice, i check them and try to mix them at some point.
frankmike lol, yeah, i get that too, its fucking weird lol, i try to fix it in the next version but im trying to learn more, before doing that. I tried a lot more progresions of the merges, with other models, but every time i gets super weird results...
Update coming soon!
"Merge your own model!" They said, "It'll be fun!" they said...
Thanks for experimenting with this. I tried to make a merge with Illustrious checkpoints, but have not been fully satisfied with the result so far.
Hi everyone, sorry for taking so long with the update, but my computer fried during a sudden power outage. I'm still waiting for the power company to pay me for the damages to everything they burned. I'll be back as soon as I can.