How to make img2img "eat" two images?

[Faishun]

Hi guys, developing Python application here. I don't really understand how to make img2img accept 2 images in post request. I get a RuntimeError: bad number of images passed: 2; expecting 1 or less.

        try:

            image_base64_list = []
            for fileName in fileNames:
                # Read the image file and encode it in base64
                with open(fileName, "rb") as image_file:
                    img_bytes = image_file.read()
                img_base64 = base64.b64encode(img_bytes).decode('utf-8')

                # Append to the image list with the expected object format
                image_base64_list.append(img_base64)

            # Create the payload
            payload = {
                "init_images": image_base64_list,
                "prompt": "beautiful house, pathway",
                "steps": 25,
                "negative_prompt": "easynegative, bad quality, worst quality, jpeg artifacts, cartoon, cgi, render, illustration, painting, drawing",
                "seed": -1,
                "sampler_name": "Euler a",
                "denoising_strength": 0.9,
                "subseed": -1,
                "batch_size": 1,
                "n_iter": 1,
                "restore_faces": False,
                "cfg_scale": 6,
                "width": 768,
                "height": 768,
                "eta": 31337,
            }

            # Handle the progress as in the single upscaling method
            self.progress_thread_running = True
            progress_thread = threading.Thread(target=self.update_progress_bar, args=(progress_signal,))
            progress_thread.start()

            # Now send the POST request
            response = requests.post(url=f'{self.url}/sdapi/v1/img2img', json=payload)

            # Tell the progress thread to stop
            self.progress_thread_running = False
            progress_thread.join()

So I'm using a list of images and get an error, however, as far as I understand, in API docs you receive an array of images and can post an array as well:

[w-e-w]

if you wish to pass multiple input images the image count and the "batch_size" needs to be equal
so if you have 2 input images you need batch size of 2

stable-diffusion-webui/modules/processing.py

Lines 1684 to 1688 in ac8ffb3

	elif len(imgs) <= self.batch_size:
	self.batch_size = len(imgs)
	batch_images = np.array(imgs)
	else:
	raise RuntimeError(f"bad number of images passed: {len(imgs)}; expecting {self.batch_size} or less")

note
it also works if the batch size is greater than the image count, as it is automatically set to the number of the input image
but that could be potentially other processes of webui or extensions that already uses the initial batch size value for something and might cause issues

also if you're working with in paint (img2img with mask) then realistically you can only pass 1 input image
not to mention most extension that takes additional image as input mostly are designed to only take 1 input image
so even if the default webui allows you to use
multiple input images, I do not recommend it

[Faishun]

Thanks, I actually didn't know about the correlation with the batch size. I don't really need inpaint, all I need is to demonstrate img2img denoising strength. I'll set batch_size to the length of input stream files.

[w-e-w]

demonstrate img2img denoising strength?

I don't know what you're doing but maybe you can just use the XYZ?

[Faishun]

Actually, no, my task is user specific, I need to make variations of the batch of images for output (users) to see. It's for educational purposes with the real world application cases.

[w-e-w]

no idea what you doing
but I guess you know what you want better than me
good luck