ControlNet-Canny: Optimized for Qualcomm Devices

On-device, high-resolution image synthesis from text and image prompts. ControlNet guides Stable-diffusion with provided input image to generate accurate images from given input prompt.

This is based on the implementation of ControlNet-Canny found here. This repository contains pre-exported model files optimized for Qualcomm® devices. You can use the Qualcomm® AI Hub Models library to export with custom configurations. More details on model performance across various devices, can be found here.

Qualcomm AI Hub Models uses Qualcomm AI Hub Workbench to compile, profile, and evaluate this model. Sign up to run these models on a hosted Qualcomm® device.

Getting Started

There are two ways to deploy this model on your device:

Option 1: Download Pre-Exported Models

Below are pre-exported model assets ready for deployment.

Runtime	Precision	Chipset	SDK Versions	Download
PRECOMPILED_QNN_ONNX	w8a16	Snapdragon® X Elite	QAIRT 2.37, ONNX Runtime 1.23.0	Download
PRECOMPILED_QNN_ONNX	w8a16	Snapdragon® 8 Gen 3 Mobile	QAIRT 2.37, ONNX Runtime 1.23.0	Download
PRECOMPILED_QNN_ONNX	w8a16	Qualcomm® QCS8550 (Proxy)	QAIRT 2.37, ONNX Runtime 1.23.0	Download
PRECOMPILED_QNN_ONNX	w8a16	Snapdragon® 8 Elite For Galaxy Mobile	QAIRT 2.37, ONNX Runtime 1.23.0	Download
PRECOMPILED_QNN_ONNX	w8a16	Snapdragon® 7 Gen 4 Mobile	QAIRT 2.37, ONNX Runtime 1.23.0	Download
PRECOMPILED_QNN_ONNX	w8a16	Snapdragon® 8 Elite Gen 5 Mobile	QAIRT 2.37, ONNX Runtime 1.23.0	Download
PRECOMPILED_QNN_ONNX	w8a16	Qualcomm® QCM6690	QAIRT 2.37, ONNX Runtime 1.23.0	Download
PRECOMPILED_QNN_ONNX	w8a16	Qualcomm® QCS9075	QAIRT 2.37, ONNX Runtime 1.23.0	Download
QNN_CONTEXT_BINARY	w8a16	Snapdragon® X Elite	QAIRT 2.42	Download
QNN_CONTEXT_BINARY	w8a16	Snapdragon® 8 Gen 3 Mobile	QAIRT 2.42	Download
QNN_CONTEXT_BINARY	w8a16	Qualcomm® QCS8275 (Proxy)	QAIRT 2.42	Download
QNN_CONTEXT_BINARY	w8a16	Qualcomm® QCS8550 (Proxy)	QAIRT 2.42	Download
QNN_CONTEXT_BINARY	w8a16	Qualcomm® SA8775P	QAIRT 2.42	Download
QNN_CONTEXT_BINARY	w8a16	Snapdragon® 8 Elite For Galaxy Mobile	QAIRT 2.42	Download
QNN_CONTEXT_BINARY	w8a16	Snapdragon® 7 Gen 4 Mobile	QAIRT 2.42	Download
QNN_CONTEXT_BINARY	w8a16	Snapdragon® 8 Elite Gen 5 Mobile	QAIRT 2.42	Download
QNN_CONTEXT_BINARY	w8a16	Qualcomm® SA7255P	QAIRT 2.42	Download
QNN_CONTEXT_BINARY	w8a16	Qualcomm® QCM6690	QAIRT 2.42	Download
QNN_CONTEXT_BINARY	w8a16	Qualcomm® QCS9075	QAIRT 2.42	Download

For more device-specific assets and performance metrics, visit ControlNet-Canny on Qualcomm® AI Hub.

Option 2: Export with Custom Configurations

Use the Qualcomm® AI Hub Models Python library to compile and export the model with your own:

• Custom weights (e.g., fine-tuned checkpoints)
• Custom input shapes
• Target device and runtime configurations

This option is ideal if you need to customize the model beyond the default configuration provided here.

See our repository for ControlNet-Canny on GitHub for usage instructions.

Model Details

Model Type: Model_use_case.image_generation

Model Stats:

• Input: Text prompt and input image as a reference
• Conditioning Input: Canny-Edge
• Text Encoder Number of parameters: 340M
• UNet Number of parameters: 865M
• VAE Decoder Number of parameters: 83M
• ControlNet Number of parameters: 361M
• Model size: 1.4GB

Performance Summary

Model	Runtime	Precision	Chipset	Inference Time (ms)	Peak Memory Range (MB)	Primary Compute Unit
controlnet	PRECOMPILED_QNN_ONNX	w8a16	Snapdragon® X Elite	59.129 ms	351 - 351 MB	NPU
controlnet	PRECOMPILED_QNN_ONNX	w8a16	Snapdragon® 8 Gen 3 Mobile	45.568 ms	33 - 45 MB	NPU
controlnet	PRECOMPILED_QNN_ONNX	w8a16	Qualcomm® QCS8550 (Proxy)	60.057 ms	30 - 31 MB	NPU
controlnet	PRECOMPILED_QNN_ONNX	w8a16	Qualcomm® QCS9075	58.171 ms	30 - 34 MB	NPU
controlnet	PRECOMPILED_QNN_ONNX	w8a16	Qualcomm® QCM6690	1040.403 ms	0 - 10 MB	NPU
controlnet	PRECOMPILED_QNN_ONNX	w8a16	Snapdragon® 8 Elite For Galaxy Mobile	33.962 ms	32 - 43 MB	NPU
controlnet	PRECOMPILED_QNN_ONNX	w8a16	Snapdragon® 7 Gen 4 Mobile	113.204 ms	33 - 39 MB	NPU
controlnet	PRECOMPILED_QNN_ONNX	w8a16	Snapdragon® 8 Elite Gen 5 Mobile	28.884 ms	31 - 42 MB	NPU
controlnet	QNN_CONTEXT_BINARY	w8a16	Snapdragon® X Elite	75.85 ms	2 - 2 MB	NPU
controlnet	QNN_CONTEXT_BINARY	w8a16	Snapdragon® 8 Gen 3 Mobile	59.518 ms	2 - 10 MB	NPU
controlnet	QNN_CONTEXT_BINARY	w8a16	Qualcomm® QCS8275 (Proxy)	139.916 ms	2 - 10 MB	NPU
controlnet	QNN_CONTEXT_BINARY	w8a16	Qualcomm® QCS8550 (Proxy)	83.656 ms	2 - 10 MB	NPU
controlnet	QNN_CONTEXT_BINARY	w8a16	Qualcomm® SA8775P	76.946 ms	2 - 11 MB	NPU
controlnet	QNN_CONTEXT_BINARY	w8a16	Qualcomm® QCS9075	75.284 ms	0 - 16 MB	NPU
controlnet	QNN_CONTEXT_BINARY	w8a16	Qualcomm® QCM6690	964.957 ms	2 - 8 MB	NPU
controlnet	QNN_CONTEXT_BINARY	w8a16	Qualcomm® SA7255P	139.916 ms	2 - 10 MB	NPU
controlnet	QNN_CONTEXT_BINARY	w8a16	Snapdragon® 8 Elite For Galaxy Mobile	46.052 ms	2 - 10 MB	NPU
controlnet	QNN_CONTEXT_BINARY	w8a16	Snapdragon® 7 Gen 4 Mobile	124.414 ms	2 - 8 MB	NPU
controlnet	QNN_CONTEXT_BINARY	w8a16	Snapdragon® 8 Elite Gen 5 Mobile	28.773 ms	2 - 11 MB	NPU
text_encoder	PRECOMPILED_QNN_ONNX	w8a16	Snapdragon® X Elite	5.624 ms	157 - 157 MB	NPU
text_encoder	PRECOMPILED_QNN_ONNX	w8a16	Snapdragon® 8 Gen 3 Mobile	3.96 ms	0 - 7 MB	NPU
text_encoder	PRECOMPILED_QNN_ONNX	w8a16	Qualcomm® QCS8550 (Proxy)	5.414 ms	0 - 162 MB	NPU
text_encoder	PRECOMPILED_QNN_ONNX	w8a16	Qualcomm® QCS9075	5.438 ms	0 - 3 MB	NPU
text_encoder	PRECOMPILED_QNN_ONNX	w8a16	Qualcomm® QCM6690	28.302 ms	0 - 6 MB	NPU
text_encoder	PRECOMPILED_QNN_ONNX	w8a16	Snapdragon® 8 Elite For Galaxy Mobile	3.167 ms	0 - 7 MB	NPU
text_encoder	PRECOMPILED_QNN_ONNX	w8a16	Snapdragon® 7 Gen 4 Mobile	5.778 ms	0 - 6 MB	NPU
text_encoder	PRECOMPILED_QNN_ONNX	w8a16	Snapdragon® 8 Elite Gen 5 Mobile	2.635 ms	0 - 10 MB	NPU
text_encoder	QNN_CONTEXT_BINARY	w8a16	Snapdragon® X Elite	3.608 ms	0 - 0 MB	NPU
text_encoder	QNN_CONTEXT_BINARY	w8a16	Snapdragon® 8 Gen 3 Mobile	2.305 ms	0 - 7 MB	NPU
text_encoder	QNN_CONTEXT_BINARY	w8a16	Qualcomm® QCS8275 (Proxy)	7.549 ms	0 - 8 MB	NPU
text_encoder	QNN_CONTEXT_BINARY	w8a16	Qualcomm® QCS8550 (Proxy)	3.239 ms	0 - 2 MB	NPU
text_encoder	QNN_CONTEXT_BINARY	w8a16	Qualcomm® SA8775P	3.505 ms	0 - 9 MB	NPU
text_encoder	QNN_CONTEXT_BINARY	w8a16	Qualcomm® QCS9075	3.304 ms	0 - 2 MB	NPU
text_encoder	QNN_CONTEXT_BINARY	w8a16	Qualcomm® QCM6690	22.025 ms	0 - 6 MB	NPU
text_encoder	QNN_CONTEXT_BINARY	w8a16	Qualcomm® SA7255P	7.549 ms	0 - 8 MB	NPU
text_encoder	QNN_CONTEXT_BINARY	w8a16	Snapdragon® 8 Elite For Galaxy Mobile	1.749 ms	0 - 13 MB	NPU
text_encoder	QNN_CONTEXT_BINARY	w8a16	Snapdragon® 7 Gen 4 Mobile	3.636 ms	0 - 7 MB	NPU
text_encoder	QNN_CONTEXT_BINARY	w8a16	Snapdragon® 8 Elite Gen 5 Mobile	1.492 ms	0 - 10 MB	NPU
unet	PRECOMPILED_QNN_ONNX	w8a16	Snapdragon® X Elite	115.766 ms	829 - 829 MB	NPU
unet	PRECOMPILED_QNN_ONNX	w8a16	Snapdragon® 8 Gen 3 Mobile	84.074 ms	13 - 26 MB	NPU
unet	PRECOMPILED_QNN_ONNX	w8a16	Qualcomm® QCS8550 (Proxy)	117.53 ms	0 - 883 MB	NPU
unet	PRECOMPILED_QNN_ONNX	w8a16	Qualcomm® QCS9075	109.17 ms	11 - 26 MB	NPU
unet	PRECOMPILED_QNN_ONNX	w8a16	Qualcomm® QCM6690	2160.0 ms	10 - 21 MB	NPU
unet	PRECOMPILED_QNN_ONNX	w8a16	Snapdragon® 8 Elite For Galaxy Mobile	67.33 ms	6 - 17 MB	NPU
unet	PRECOMPILED_QNN_ONNX	w8a16	Snapdragon® 7 Gen 4 Mobile	177.928 ms	9 - 19 MB	NPU
unet	PRECOMPILED_QNN_ONNX	w8a16	Snapdragon® 8 Elite Gen 5 Mobile	46.028 ms	13 - 23 MB	NPU
unet	QNN_CONTEXT_BINARY	w8a16	Snapdragon® X Elite	113.746 ms	13 - 13 MB	NPU
unet	QNN_CONTEXT_BINARY	w8a16	Snapdragon® 8 Gen 3 Mobile	80.577 ms	13 - 20 MB	NPU
unet	QNN_CONTEXT_BINARY	w8a16	Qualcomm® QCS8275 (Proxy)	264.31 ms	8 - 15 MB	NPU
unet	QNN_CONTEXT_BINARY	w8a16	Qualcomm® QCS8550 (Proxy)	112.347 ms	10 - 12 MB	NPU
unet	QNN_CONTEXT_BINARY	w8a16	Qualcomm® SA8775P	106.976 ms	10 - 17 MB	NPU
unet	QNN_CONTEXT_BINARY	w8a16	Qualcomm® QCS9075	106.531 ms	11 - 25 MB	NPU
unet	QNN_CONTEXT_BINARY	w8a16	Qualcomm® QCM6690	2039.478 ms	13 - 24 MB	NPU
unet	QNN_CONTEXT_BINARY	w8a16	Qualcomm® SA7255P	264.31 ms	8 - 15 MB	NPU
unet	QNN_CONTEXT_BINARY	w8a16	Snapdragon® 8 Elite For Galaxy Mobile	61.075 ms	6 - 18 MB	NPU
unet	QNN_CONTEXT_BINARY	w8a16	Snapdragon® 7 Gen 4 Mobile	179.697 ms	10 - 15 MB	NPU
unet	QNN_CONTEXT_BINARY	w8a16	Snapdragon® 8 Elite Gen 5 Mobile	47.215 ms	13 - 22 MB	NPU
vae	PRECOMPILED_QNN_ONNX	w8a16	Snapdragon® X Elite	219.597 ms	59 - 59 MB	NPU
vae	PRECOMPILED_QNN_ONNX	w8a16	Snapdragon® 8 Gen 3 Mobile	163.642 ms	4 - 11 MB	NPU
vae	PRECOMPILED_QNN_ONNX	w8a16	Qualcomm® QCS8550 (Proxy)	219.435 ms	0 - 66 MB	NPU
vae	PRECOMPILED_QNN_ONNX	w8a16	Qualcomm® QCS9075	207.133 ms	3 - 6 MB	NPU
vae	PRECOMPILED_QNN_ONNX	w8a16	Qualcomm® QCM6690	5359.841 ms	0 - 7 MB	NPU
vae	PRECOMPILED_QNN_ONNX	w8a16	Snapdragon® 8 Elite For Galaxy Mobile	147.795 ms	3 - 16 MB	NPU
vae	PRECOMPILED_QNN_ONNX	w8a16	Snapdragon® 7 Gen 4 Mobile	445.729 ms	3 - 9 MB	NPU
vae	PRECOMPILED_QNN_ONNX	w8a16	Snapdragon® 8 Elite Gen 5 Mobile	89.515 ms	3 - 14 MB	NPU
vae	QNN_CONTEXT_BINARY	w8a16	Snapdragon® X Elite	266.163 ms	0 - 0 MB	NPU
vae	QNN_CONTEXT_BINARY	w8a16	Snapdragon® 8 Gen 3 Mobile	202.114 ms	0 - 8 MB	NPU
vae	QNN_CONTEXT_BINARY	w8a16	Qualcomm® QCS8275 (Proxy)	725.838 ms	0 - 9 MB	NPU
vae	QNN_CONTEXT_BINARY	w8a16	Qualcomm® QCS8550 (Proxy)	273.289 ms	1 - 5 MB	NPU
vae	QNN_CONTEXT_BINARY	w8a16	Qualcomm® SA8775P	1136.64 ms	0 - 9 MB	NPU
vae	QNN_CONTEXT_BINARY	w8a16	Qualcomm® QCS9075	249.997 ms	0 - 3 MB	NPU
vae	QNN_CONTEXT_BINARY	w8a16	Qualcomm® QCM6690	6374.38 ms	0 - 6 MB	NPU
vae	QNN_CONTEXT_BINARY	w8a16	Qualcomm® SA7255P	725.838 ms	0 - 9 MB	NPU
vae	QNN_CONTEXT_BINARY	w8a16	Snapdragon® 8 Elite For Galaxy Mobile	178.859 ms	0 - 13 MB	NPU
vae	QNN_CONTEXT_BINARY	w8a16	Snapdragon® 7 Gen 4 Mobile	512.932 ms	0 - 6 MB	NPU
vae	QNN_CONTEXT_BINARY	w8a16	Snapdragon® 8 Elite Gen 5 Mobile	116.219 ms	0 - 9 MB	NPU

License

• The license for the original implementation of ControlNet-Canny can be found here.

References

• Adding Conditional Control to Text-to-Image Diffusion Models
• Source Model Implementation

Community

• Join our AI Hub Slack community to collaborate, post questions and learn more about on-device AI.
• For questions or feedback please reach out to us.

Usage and Limitations

This model may not be used for or in connection with any of the following applications:

• Accessing essential private and public services and benefits;
• Administration of justice and democratic processes;
• Assessing or recognizing the emotional state of a person;
• Biometric and biometrics-based systems, including categorization of persons based on sensitive characteristics;
• Education and vocational training;
• Employment and workers management;
• Exploitation of the vulnerabilities of persons resulting in harmful behavior;
• General purpose social scoring;
• Law enforcement;
• Management and operation of critical infrastructure;
• Migration, asylum and border control management;
• Predictive policing;
• Real-time remote biometric identification in public spaces;
• Recommender systems of social media platforms;
• Scraping of facial images (from the internet or otherwise); and/or
• Subliminal manipulation