Weird results using API

[dayunbao]

I'm getting very different results when using the API vs generating an image using the web UI.

The web UI values:

{
    "alwayson_scripts": {
        "ControlNet": {
            "args": [
                {
                    "advanced_weighting": null,
                    "animatediff_batch": false,
                    "batch_image_files": [],
                    "batch_images": "",
                    "batch_keyframe_idx": null,
                    "batch_mask_dir": null,
                    "batch_modifiers": [],
                    "control_mode": "Balanced",
                    "effective_region_mask": null,
                    "enabled": true,
                    "guidance_end": 1.0,
                    "guidance_start": 0.0,
                    "hr_option": "Both",
                    "image": {
                        "image": "base64image placeholder",
                        "mask": "base64image placeholder"
                    },
                    "inpaint_crop_input_image": false,
                    "input_mode": "simple",
                    "ipadapter_input": null,
                    "is_ui": true,
                    "loopback": false,
                    "low_vram": true,
                    "mask": null,
                    "model": "controlnet-canny-sdxl-1.0 [acfd43c3]",
                    "module": "canny",
                    "output_dir": "",
                    "pixel_perfect": true,
                    "processor_res": 512,
                    "pulid_mode": "Fidelity",
                    "resize_mode": "Just Resize",
                    "save_detected_map": true,
                    "threshold_a": 150.0,
                    "threshold_b": 10.0,
                    "weight": 1.0
                }
            ]
        }
    },
    "batch_size": 4,
    "cfg_scale": 7,
    "comments": {},
    "denoising_strength": 0.2,
    "disable_extra_networks": false,
    "do_not_save_grid": false,
    "do_not_save_samples": false,
    "enable_hr": false,
    "height": 1024,
    "hr_negative_prompt": "",
    "hr_prompt": "",
    "hr_resize_x": 0,
    "hr_resize_y": 0,
    "hr_scale": 2,
    "hr_second_pass_steps": 10,
    "hr_upscaler": "R-ESRGAN 4x+",
    "n_iter": 1,
    "negative_prompt": " bad_prompt_version2-neg,  easynegative",
    "override_settings": {},
    "override_settings_restore_afterwards": true,
    "prompt": " A cyberpunk bar <lora:Cyberpunk _Anime_V2_light_version-SDXL:0.7>",
    "restore_faces": false,
    "s_churn": 0.0,
    "s_min_uncond": 0,
    "s_noise": 1.0,
    "s_tmax": null,
    "s_tmin": 0.0,
    "sampler_name": "DPM++ 2M",
    "scheduler": "Automatic",
    "script_args": [],
    "script_name": null,
    "seed": -1,
    "seed_enable_extras": true,
    "seed_resize_from_h": -1,
    "seed_resize_from_w": -1,
    "steps": 50,
    "styles": [],
    "subseed": -1,
    "subseed_strength": 0,
    "tiling": false,
    "width": 1024
}

Input image:

Result:

However, when I do the same thing using the API, I get results like this:
Input image:

Result:

The ControlNet detect request body:

 control_net_detect_request_body = {
        'controlnet_module': 'canny',
        'controlnet_input_images': [image_str],
        'controlnet_processor_res': image_size['width'] if image_size['width'] > image_size['height'] else image_size[
            'height'],
        'controlnet_threshold_a': 150,
        'controlnet_threshold_b': 10,
    }

The text2img request body:

control_net_settings = {
        'enabled': True,
        'image': image_str,
        'low_vram': True,
        'pixel_perfect': True,
        'module': "canny",
        # SD XL version
        'model': 'controlnet-canny-sdxl-1.0 [acfd43c3]',
        'weight': 1.0,
        'guidance_start': 0.0,
        'guidance_end': 1.0,
        'control_mode': 'Balanced',
        'resize_mode': 'Just Resize',
        'processor_res': image_size['width'],
        'threshold_a': 150,
        'threshold_b': 10,
    }

    text_to_image_request_body = {
        'prompt': background_prompt,
        'negative_prompt': "bad_prompt_version2-neg, easynegative",
        'width': image_size['width'],
        'height': image_size['height'],
        'cfg_scale': 7,
        'sampler_name': "DPM++ 2M",
        'scheduler': 'Automatic',
        'batch_size': 1,
        'seed': -1,
        'steps': 50,
        'alwayson_scripts': {
            'controlnet': {
                'args': [control_net_settings]
            },
        }
    }

The image is blurry and garishly colored. Am I doing something wrong?

[dayunbao]

Never mind. I figured it out. The size of the image was wonky.