Ram Charan is an Indian film actor and producer who works in the Telugu film industry. He was born on March 27, 1985 in Chennai, Tamil Nadu, India. He is the son of the legendary Telugu actor, Chiranjeevi. Charan made his acting debut in 2007 with the film "Chirutha". He has since appeared in numerous successful films and has established himself as one of the leading actors in Telugu cinema.
Charan is known for his versatile acting skills and impressive dancing abilities and is considered one of the best performers in the Telugu film industry. He is also famous for his good looks and charismatic on-screen presence. Some of his most popular films include "Magadheera" (2009), "Yevadu" (2014), and "Rangasthalam" (2018). In addition to acting, Charan has also been involved in the production of several successful films and has won several awards and accolades for his contributions to Telugu cinema.
Charan is most famous for his acting career, particularly for his roles in successful Telugu films and for his dancing abilities. He is widely popular among Telugu film fans and is considered a leading figure in the Telugu film industry. His recent movie RRR has become a worldwide sensation.
Description
FAQ
Comments (11)
You dont need to make a checkpoint for one person's face, a pickle tensor suffices
what is pickle tensor?
besides DB I know of TI, Lora and Hypernetwork
since I can't make hypernetworks and Lora/TI cannot get 99.99% similarity with the original I'm making DBs currently
If you want to help, you could convert it to Lora so that people interested in Ram Charan (that do not have enought hdd space) could also use it :)
@malcolmrey I dont know how to use LoRA, if you want to make an embedding follow this guide https://www.youtube.com/watch?v=2ityl_dNRNw&t=33s it saves as a pt file pickle is a python library for sequencing and unsequencing data (unpickling they call it)
A tensor is complex 3d array that holds a lot of info (its name cames from cramming so much info in a small space so tensely its going to explode) it would be literally a million+ times smaller, I cant spare the compute units to make it :(
@SpacePatrice i feel you, I was able to configure dreambooth for myself but I fail to run hypernetworks :-(
I've got some tips how to run TI but haven't had time yet to see if it works for me.
BTW, I saw your TIs - they are nice but I'm wondering if you were able to generate an output that would fool someone into thinking that they are looking at an actual photo?
Since you've provided training data I'm eager to make a dreambooth just to compare the results.
I'll also include training data for Ram Charan so if you you are able to do a TI for comparison sake - that would be great.
Actually the perfect scenario for me is that for a given person there is always a dreambooth, TI, hypernetwork and Lora so that whoever wants to use it - can pick what they actually want/need :)
check out my Milly Alcock Review https://civitai.com/models/4388/milly-alcock-textual-embedding pics they're very realistic with 1.2
I edit/use a prompt I found here https://www.reddit.com/r/StableDiffusion/comments/10lodi0/women_on_water_realistic_vision_12/
high angle shot (kneeling and bending over on water wearing grey blue micro top and hot shorts:1.23) (short wavy hair:1.2), gentle smile ray traced shadows, RAW, 8k, (eczema:0.7), (sub-surface scattering:1.55), (sweat:1.22), (freckles:0.55), highly detailed skin, (Acne:0.7), (FACE, perfect eyes, no makeup. (skin spores:1.05), (skin spores:1.05), ultra detailed face, ultra detailed skin, film grain, ray tracing, studio lighting
Negative prompt: signature, watermark, airbrush, photoshop, plastic doll, (ugly eyes, deformed iris, deformed pupils, fused lips and teeth:1.2), (un-detailed skin, semi-realistic, cgi, 3d, render, sketch, cartoon, drawing, anime:1.2), text, close up, cropped, out of frame, worst quality, low quality, jpeg artifacts, ugly, duplicate, morbid, mutilated, extra fingers, mutated hands, poorly drawn hands, poorly drawn face, mutation, deformed, blurry, dehydrated, bad anatomy, bad proportions, extra limbs, cloned face, disfigured, gross proportions, malformed limbs, missing arms, missing legs, extra arms, extra legs, fused fingers, too many fingers, long neck, head wear, masculine, obese, fat, out of frame
Steps: 20
Sampler: DPM++ 2M Karras / DPM ++ SDE Karras
CFG scale: 7
@malcolmrey
@SpacePatrice I've been trying to train embeddings with no luck. For a set of 20-30 photos, which config. do you recommend?
I mean steps, batch size, and gradient accumulation.
@elen7 copy and paste 0.05:10, 0.02:20, 0.01:60, 0.005:200, 0.002:500, 0.001:3000, 0.0005 for the steps and follow the guidelines from this https://www.youtube.com/watch?v=2ityl_dNRNw&t=33s
the images need to be high fidelity (upscaling works if all images are bad) and 512x512 with description you can generate with BLIP and double check , token size is 2-3 for 10-20 pics 4-6 for 20-30 you can even go as high as 8 because the closer to 10 tokens the better for faces
make sure the setting are changed too like the video shows, ig thats it
You can train faces, and the size of the file will be just a few kilobytes or MB, and you will still generate a good face. Like Textual Inversion, Lora etc.
By the way, are you a Telugu guy?
@bhagathgoud i will eventually extract it into LORA so the filesize will be smaller
I do not like TI for faces as they do not produce the BEST results.
Nope, I am from Europe. But I love RRR. Congracts on the Oscar for the best song! :)
can you do more Indian celebrities please, here is data set for one https://imgur.com/a/e1XnM07 if face not clear please take few more from internet and is there any other way to contact you like reddit or twitter?
Hey hey! For sure, the dataset looks promising so I'll give it a try :-)
this is Tamanna Bhatia, right?
my Reddit handle is: https://www.reddit.com/user/malcolmrey
Looks like we don't have an active mirror for this file right now.
CivArchive is a community-maintained index — we catalog mirrors that volunteers upload to HuggingFace, torrents, and other public hosts. Looks like no one has uploaded a copy of this file yet.
Some files do get recovered over time through contributions. If you're looking for this one, feel free to ask in Discord, or help preserve it if you have a copy.
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.