UPDATE 11/02/2026: FOR THE QWEN-IMAGE-2512 VERSION, SEE HERE: [Qwen-Image-2512] Smartphone Snapshot Photo Reality [STYLE] - v10.0 | Qwen LoRA | Civitai
Donations: Training these new heavy models like Qwen-Image and WAN2.2 is very expensive for me and I don't earn any money with these models, so any donation is appreciated to keep my models free and fund training new ones: https://ko-fi.com/aicharacters
Trigger (must include in prompt or else LoRa likeness will be very lacking):
amateur photo
Please note that the online generator will NOT give you as good results as running the model locally in ComfyUI using my following recommended inference workflow: https://www.dropbox.com/scl/fi/u5x0aehj9qvumx0uyb55c/Qwen-Image_recommended_default_text2image_inference_workflow_by_AI_Characters.json?rlkey=8xf1fian7xcoxpckswq7f8ip9&st=bwijiu0a&dl=1
Latest version notes and model page updates:
initial release
LoRa description:
This LoRa model was designed to generate images that look like casual snapshot photos taken with a smartphone.
Disclaimer:
I do not claim to be the current owner, original creator of, person or legal representative of a person depicted in, any of the concepts that my models, including this one, aim to emulate nor the training data used to train this model. I do not claim that any of my models have received an endorsement by those respective individuals. I also do not claim that the emulation attempted by my models represents a 100% accurate depiction of the original concept in question nor do I claim that the quality of my emulation reaches the same quality of the original concept. All credit goes to the respective current owners, original creators, or people and I encourage you to support them in any way you can.
If you are the current owner, original creator of, person or legal representative of a person depicted in, any of the concepts that any of my models aims to emulate and would like that specific model to be removed, then please leave me a private message here or on Reddit with proof of authenticity of your identity and claim and I will remove it.
Additionally, I do not endorse my models being used to violate the law or enact still legal, but immoral acts such as the spread of misinformation using deep fakes.
Description
FAQ
Comments (50)
Why are your sample images using the same prompts as Danrisi's SamsungCam UltraReal Qwen lora?
cuz why not - it's actually a pretty great way to compare realism models - to use familiar prompts you've seen elsewhere to see what the model is capable of. I like it.
It's our old game since my old flux releases. We take each other's prompts for lora showcases
@Danrisi Ah, that makes sense. I was very confused at first, because I knew I had seen that Cetaphil/Tanqueray image before, and I didn't know if you had made a new account or if there was something shady going on.
imo that workflow isn't ideal, 40 steps AND a multistep sampler (res_2s) is completely overkill. it takes nearly 10 minutes to generate an image on a 3090 with that lmao
Youre free to use whatever workflow you like. I merely shared what is the optimal workflow for this model for the best possible combination of coherence, prompt adherence, and quality.
@AI_Characters how dare you give out a free model and not a workflow that is perfect for everyone else and not yourself? sheeeesh. get it together buddy
20 step on euler here, works fantastic xd
nunchaku bro is edging us with the lora support tho
can you train a nsfw qwen lora? please
I will tell you a trick. Find an image that the style, quality, color.. everything, highest res you can find with a similar body.
Use the (2509?).
Place the target hi res great image of a similar person as image 1
Place the person to transform as image 2
This is a prompt:
"Using the nude woman in image 1 and the nude woman from image two, make a new scene of a slate grey backdrop. Two nude women are standing in front of a slate grey backdrop with their arms relaxed near their hips, staring at the viewer."
I have tried this with a person I know who was amazed to see how closely it rendered her nudity without a reference. Scary close.
Qwen will surprise you if you know how to find the edges.
This is great! Thank you!
Can it do NSFW images to?
They should be the same as default Qwen.
Excellent work. Like other pointed out in reddit, when it likes the prompt, it works very well. If it doesn't like the prompt, the default qwen look creeps in (Maybe not the default qwen look but it looks a bit like base Flux imo).
A couple of things I noticed are it struggles to generate low light shots. Either it makes it bright or it wants to use camera flash. It also likes to add freckles (Not like how Flux Krea adds it but it does add them without me asking for it).
Aside from that I think you did an amazing job with this Lora. Its 1000 times better than all your other loras haha :)
Sorry to pry, about how much does it cost to make a lora like this?
I ask because I will add credits soon and I owe you for this one!
Looks great.
I did not train this using the CivitAI trainer. I trained this manually using a rented GPU. H200 on Vast.ai 2.5€/h.
@AI_Characters thank you for the work, it looks great..
but my Comfy installation just broke! lol I followed you, all of your stuff looks excellent!
@AI_Characters thank you for you work! cant wait to see updates from you or new loras, this one looks awesome!
Works freaking great, thx a lot good sir.
Thank you for your excellent work! Although, if you ignore the beautiful dog, 99% of the time your Lora creates women. Although the tricks with 2509 work well; creating women dressed, half-dressed, and naked at the same time is much more exciting. Unfortunately, if you add NSWF Lora to Lora, everything falls apart – all the magical air, small details, the play of light and shadow disappears. If suddenly, someday, you could add some genitals to her – that would be the crown jewel of QWEN :) So that QWEN would simply know what's there, without any poses or interactions.
I dont know what you mean. You can generate male men just fine. You just need to prompt that.
Try blocking some layers of the NSFW lora, LoRA Loader (Block Weight) in ComfyUI
@AI_Characters I'm sorry, I don't speak English very well, I had a guide translate the text. If you could teach LORA to understand, say, hairy pussy. This would allow for a full-fledged generation of ordinary to erotic images.
Found a suitable NSWF LORA https://civitai.com/models/2020014/qwennsfwbody
@alezzby that NSFW lora from your link looks pretty bad lol you're not difficult I guess xDD
give me tips please using with 2509 , do you know how to prevent body change while do change clothes (try-on) , is generate different body ..
有一个建议,不知道对不对
可以申请支付宝
很多中国人玩ai的
many Chinese play ai? lol. C'mon qwen image itself is created by a Chinese company. of course Chinese play ai. you should say many Chinese play civitai
人家老外说的对 这个AI就是国产的..
I've tried prompt like in examples and very similar results - incredible details.
Absolutely amazing lora!!! Thanks!
Please tell me you are working on a Z-Image port of this! Been using this for all my Qwen generations! Incredible work and realism!
Amazing!! Is there a Qwen-Image 2512 version?
I am working on it.
@AI_Characters you're the best!
Still interested in a 2512!
@fohowop124538fdfdfdfds I am still working on it. As I am also on a ZTurbo, Base and Klein Base 9b version. Release should be soon though.
There you go: [Qwen-Image-2512] Smartphone Snapshot Photo Reality [STYLE] - v10.0 | Qwen LoRA | Civitai
@fohowop124538fdfdfdfds
does it not work with qwen-edit 2511 ? I don't see the difference with and without Lora
It wasn't trained for Edit.
@AI_Characters Just FYI - I am using it with edit and it's working perfectly! Maybe he forgot to put amateur photo in the prompt.
The world would be better with a a 2512 version.
I am still working on it. As I am also on a ZTurbo, Base and Klein Base 9b version. Release should be soon though.
The Qwen-Image-2512 version can be found here: [Qwen-Image-2512] Smartphone Snapshot Photo Reality [STYLE] - v10.0 | Qwen LoRA | Civitai
can u use this for img2img? :/
Details
Files
Qwen-Image_SmartphoneSnapshotPhotoReality_v4_by-AI_Characters_TRIGGER$amateur photo$.safetensors
Mirrors
Qwen-Image_SmartphoneSnapshotPhotoReality_v4_by-AI_Characters_TRIGGER$amateur photo$.safetensors
g2-qwen-2.safetensors
Qwen-Image_SmartphoneSnapshotPhotoReality_v4_by-AI_Characters_TRIGGER$amateur photo$.safetensors
Qwen-Image_SmartphoneSnapshotPhotoReality_v4_by-AI_Characters_TRIGGER$amateur photo$.safetensors
Qwen-Image_SmartphoneSnapshotPhotoReality_v4_by-AI_Characters_TRIGGER$amateur photo$.safetensors
Qwen-Image_SmartphoneSnapshotPhotoReality_v4_by-AI_Characters_TRIGGER$amateur photo$.safetensors
Qwen-Image_SmartphoneSnapshotPhotoReality_v4_by-AI_Characters_TRIGGER$amateur photo$.safetensors
Qwen-Image_SmartphoneSnapshotPhotoReality_v4_by-AI_Characters_TRIGGER$amateur photo$.safetensors
Qwen-Image_SmartphoneSnapshotPhotoReality_v4_by-AI_Characters_TRIGGER$amateur photo$.safetensors
Qwen-Image_SmartphonePhotoReality_v4_TRIGGERamateur photo.safetensors
Qwen_Smartphone_v4_amateur photo.safetensors
Qwen-Image_Smartphone.safetensors
Qwen-Image_SmartphoneSnapshotPhotoReality_v4_by-AI_Characters_TRIGGER$amateur photo$.safetensors
Qwen-Image_SmartphoneSnapshotPhotoReality_v4_by-AI_Characters_TRIGGER$amateur photo$.safetensors
Qwen-Image_SmartphoneSnapshotPhotoReality_v4_by-AI_Characters_TRIGGER$amateur photo$.safetensors
Qwen-Image_SmartphoneSnapshotPhotoReality_v4_by-AI_Characters_TRIGGER$amateur photo$.safetensors
Qwen-Image_SmartphoneSnapshotPhotoReality_v4_by-AI_Characters_TRIGGER$amateur photo$.safetensors
Qwen-Image_SmartphoneSnapshotPhotoReality_v4_by-AI_Characters_TRIGGER$amateur photo$.safetensors
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.



















