Try my new Checkpoint for much better Quality and additional concepts
TgirlsXL
TgirlsXL is a step up from the SD1.5 version in realism and reliablility
Can create flaccid penises in almost every pose
follows prompts much better (large/small or flaccid/erect)
better looking penises
much better at side poses and poses from behind
produces much less body horror like multiple dicks...
The penis bulge concept from V4 is not working in this version
Works great in combination with my other XL Lora
Planned improvements:
Uncut penises
Tgirl on Tgirl and Tgirl on Woman poses
The Model has been trained on default SDXL and is primarily intended for realistic models but I will probably also work for anime models.
General Tips:
The recommended strength is 0.5-0.9, a higher strengths sometimes creates more realistic results.
The model seems to follow size and state (erect/flaccid) better if they come close to the beginning of the prompt
size and state (erect/flaccid) works better if the prompt is shorter
Use highres fix!!! The upsampler does not really matter. Denoise should be at least 0.3.
Use the regular model for impainting penises. Select "only mask" set Padding to a >50px and set denoiseing to ~0.6
Look at the samples for prompts
Description
New Version based on SDXL. Rendering quality and realism are much improved. It also creates way less body horror like multiple dicks compared to V4
FAQ
Comments (44)
Finally, some good fucking food. <3
ty!! would you make an extra big size variant? =P
"Large erect penis" usually already creates very large dicks. How big are we taking about
@ffjggrtbjibv Like Mandingo or bigger... For the futa community, we're used to them being pretty oversized :P
Will there be a BetterCocksXL, and will it be exactly the same as TGirlsXL?
Please consider making a Lora version for xl. Diffusers doesn't support lycoris therefore users who implement SD that way cannot use this. It would be greatly appreciated
Doesn't work your XL model in SD.
The XL version is absolutely EXCELLENT at creating visually correct girlcock every single time. My only complaint is that it seems to have a very high effect on the overall aesthetic of the image, at least on Dreamshaper. Will try some other models.
everything i gen with this turns orange and completely baked. I'm not sure why.
This happened to me too sometimes. It seems like a bug restarting the webui fixes it for me.
Lower the cfg
@ffjggrtbjibv my issue here was the lora weight and also the checkpoint model. both issues fixed by mimicing lora weight in your photos (.5) and using a different checkpoint model. Nice work!
Will XL version get a penis bulge at some point? <3
@Juice69 I can't get these additional concepts to reliably work in a single lora, so there will be different loras, I plan to make one for bulges and one for uncut dicks
Hey ,it's strange,the XL model doesnt show up in my loras (auto1111), am i the only one ???
for me it showed before i installed webui Forge now i get,
"activating extra network lora with arguments [<modules.extra_networks.ExtraNetworkParams object at 0x0000020C60A27970>, <modules.extra_networks.ExtraNetworkParams object at 0x0000020CA606D840>]: AttributeError
Traceback (most recent call last):
File "D:\Utilities\Webui\stable-diffusion-webui\modules\extra_networks.py", line 135, in activate
extra_network.activate(p, extra_network_args)
File "D:\Utilities\Webui\stable-diffusion-webui\extensions-builtin\Lora\extra_networks_lora.py", line 43, in activate
networks.load_networks(names, te_multipliers, unet_multipliers, dyn_dims)
File "D:\Utilities\Webui\stable-diffusion-webui\extensions-builtin\Lora\networks.py", line 51, in load_networks
compiled_lora_targets.append([a.filename, b, c])
AttributeError: 'NoneType' object has no attribute 'filename' "
There's either no dick or it completely changes the person. Looks like faces/style is baked into it.
For example, if you type the name of a well kown star you can easily make without the lora and then add it I tried anything between .5 and 1 and it's either no dick (or cursed) or with a dick but the face is changed to something else, most of the time both.
Using XL
is posting limited here? I have tried to post a few things but most of them don’t show up. I posted one set that was 4 images but only 1 showed up, other times nothing showed up at all.
I think there is some kind of bug on civitai. Your generations are amazing btw sucks that they don't show up here
@ffjggrtbjibv Darn. Would love to help build out this gallery. This lora is **incredible**. Super realistic, super reliable. Leaps and bounds over the 1.5 model. It fights the style of non-realistic a little bit, but easy to workaround with inpaint. Most of my results are straight out of txt2img, which I was never really happy with using 1.5 models. Keep up the good work!
@secretivethoughts have you tried posting the images by clicking "Add Post" on this model page? This should work even if the auto detection is not.
@ffjggrtbjibv yeah I have been posting from the tgirls page with add post. The one with the jungle warrior was posted as a set using that button, but only the first one appeared.
@ffjggrtbjibv, do you have a discord? I'd love to hire you. Please let me know
I'm bonscott2906 on discord
request sent! Thanks
most penis generations are slightly blurry. any fix for this?
I have not really noticed this, but you could try adetailer with this detection model https://civitai.com/models/132388?modelVersionId=149639, also try adding low resolution to the nagative prompt
3 things,
1 this is the most stable futa lora there is, great job!
2 it can't really do black tgirls, it turns them into latinas, i understand this might be very hard to fix
3 really looking forward to tgirl on tgirl or tgril on woman poses ;D
Hi. Many thanks for your Loras. I’ve had a good time with them. Could I ask if you can share the settings of your training please? How many images? What kind of them? How do you prefer to tag them? If you don’t mind. I’d like to train a Lora of small dicks
Not working with after installing Webui Forge every other lora seems to show up but not tgirl not showing up even when extra networks option is ticked, now i get
activating extra network lora with arguments [<modules.extra_networks.ExtraNetworkParams object at 0x0000020C60A27970>, <modules.extra_networks.ExtraNetworkParams object at 0x0000020CA606D840>]: AttributeError
Traceback (most recent call last):
File "D:\Utilities\Webui\stable-diffusion-webui\modules\extra_networks.py", line 135, in activate
extra_network.activate(p, extra_network_args)
File "D:\Utilities\Webui\stable-diffusion-webui\extensions-builtin\Lora\extra_networks_lora.py", line 43, in activate
networks.load_networks(names, te_multipliers, unet_multipliers, dyn_dims)
File "D:\Utilities\Webui\stable-diffusion-webui\extensions-builtin\Lora\networks.py", line 51, in load_networks
compiled_lora_targets.append([a.filename, b, c])
AttributeError: 'NoneType' object has no attribute 'filename'
Is there any way you could make a lora version of this? I'm trying to make women with dicks using self hostable software only. Lycoris isnt compatible with anything. Would be nice if there was a penis tgirl lora xl model somewhere.
sure wish they'd do this, diffusers doesn't support lycoris entirely
by using lycoris they cut off a lot of people who can use their model
Sorry but there probably won't be a Lora version. I tried to train one multiple times with different settings. The results were always really bad, wasted like 80 dollars in cloud computing costs trying
lycoris works in Fooocus local install. You just need to put it lora folder and use it as lora
This is working pretty amazing!
Love this Lora but cannot get the penis outline to work at all, any tips? SD1.5 version. Thanks!
Can i use this with a negative weight to prevent penis on a female when trying to create straight porn?
No, you get really strange results if you do that
when using XL version i get only medium shots. No way to get full body
How about sitting her down, paint her toenails, give some foot accessories like shoes or toelings? Adding prompts about missing body parts may helps you, I guess.
I put the TgirlsXL.safetensors file in the Lora folder but it doesn't show up in the WebUI. The tgirls_v4 does show up, however. Can someone help? Are there some extra steps that I missed?
Did you check your base model is for SDXL? I guess your environment is Automatic1111 and you've chose SD1.5 model. If so, incompatible version lora(TgirlsXL.safetensors) is hidden automatically, and correct version lora(tgirls_v4) is shown.
@riorio I see. Now it appears in the lora menu. Thank you for the answer!
Hi can somebody please help me trouble shoot why it isnt working with comfyui, I always get the error but couldnt find how to fix it, the model worked with automatic11111 ui fine Error occurred when executing CheckpointLoaderSimple: ERROR: Could not detect model type of: L:\Apps\checkpoints\TgirlsXL.safetensors File "L:\Apps\ComfyUi\ComfyUI_windows_portable\ComfyUI\execution.py", line 152, in recursive_execute output_data, output_ui = get_output_data(obj, input_data_all) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "L:\Apps\ComfyUi\ComfyUI_windows_portable\ComfyUI\execution.py", line 82, in get_output_data return_values = map_node_over_list(obj, input_data_all, obj.FUNCTION, allow_interrupt=True) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "L:\Apps\ComfyUi\ComfyUI_windows_portable\ComfyUI\execution.py", line 75, in map_node_over_list results.append(getattr(obj, func)(**slice_dict(input_data_all, i))) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "L:\Apps\ComfyUi\ComfyUI_windows_portable\ComfyUI\nodes.py", line 518, in load_checkpoint out = comfy.sd.load_checkpoint_guess_config(ckpt_path, output_vae=True, output_clip=True, embedding_directory=folder_paths.get_folder_paths("embeddings")) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "L:\Apps\ComfyUi\ComfyUI_windows_portable\ComfyUI\comfy\sd.py", line 517, in load_checkpoint_guess_config raise RuntimeError("ERROR: Could not detect model type of: {}".format(ckpt_path))
I ABSOLUTELY LOVE THIS MODEL, you made a perfect work
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.