def C16(prompt_data, img): # 图生图 prompt_data["22"]["inputs"]["image"] = img[0] # 替换关键词 # prompt_data["331"]["inputs"]["text"] = prompt return prompt_data def C17(prompt_data, img): # 上传姿势 # random_number = random.randint(1, 9) # 9张姿势图 # input_image_pose = './Image/C17/' + str(random_number) + '.png' # 图生图 input_image_pose = './Image/C17/1.jpg' with open(input_image_pose, "rb") as f: imgPose = upload_file(server_address, f, "", True) prompt_data["331"]["inputs"]["image"] = imgPose # 姿势图,现在应该是默认写死在json文件中 prompt_data["432"]["inputs"]["image"] = img[0] # 用户的输入图 # 指定数字长度，例如16位 # number_length = 16 # random_number = generate_random_number(number_length) # prompt_data["354"]["inputs"]["seed"] = random_number # random_number = generate_random_number(number_length) # prompt_data["425"]["inputs"]["seed"] = random_number # 替换关键词 # prompt_data["331"]["inputs"]["text"] = prompt return prompt_data def C19(prompt_data, img): # prompt_data["12"]["inputs"]["image"] = img[0] if len(img) > 1: prompt_data["27"]["inputs"]["image"] = img[1] else: prompt_data["27"]["inputs"]["image"] = img[0] return prompt_data def C20(prompt_data, img): # prompt_data["12"]["inputs"]["image"] = img[0] if len(img) > 1: prompt_data["27"]["inputs"]["image"] = img[1] else: input_image_style = './Image/C20/meixi.png' # 只上传一张就与梅西握手 with open(input_image_style, "rb") as f: image_style = upload_file(server_address, f, "", True) prompt_data["27"]["inputs"]["image"] = image_style return prompt_data def C22(prompt_data, img, model_id): # TODO 这里的逻辑应该不对了，需要在重复检查22号模型时调整代码 ####------------------------------------------------------------#### # 记录下用户的风格图+姿势图的组合，下次生成时，尽时避开已经输出过的组合，让用户更好的玩耍 # 获取目录下有多少个pose,多少个style directory_path = './Image/C' + str(model_id) + '/' pose_count = getPrefixFileCount(directory_path, 'pose_') style_count = getPrefixFileCount(directory_path, 'style_') style_id = random.randint(1, style_count) # 查找未使用过的姿态图与姿势图，如果找不到就随机用一个 pose_id = random.randint(1, max(pose_count, 1)) # for i in range(1, style_count + 1): # for j in range(1, pose_count + 1): # if not checkStylePoseIdExists(user_id, model_id, i, j): # style_id = i # pose_id = j # 回写数据库表，标识这些组合已经用过 # record_style_pose_id(user_id, model_id, style_id, pose_id) ####------------------------------------------------------------#### # 风格图 input_image_style = './Image/C' + str(model_id) + '/style_' + str(style_id) + '.png' with open(input_image_style, "rb") as f: image_style = upload_file(server_address, f, "", True) prompt_data["118"]["inputs"]["image"] = image_style # 样式图 input_image_pose = './Image/C' + str(model_id) + '/pose_' + str(pose_id) + '.png' with open(input_image_pose, "rb") as f: image_pose = upload_file(server_address, f, "", True) prompt_data["98"]["inputs"]["image"] = image_pose # 原图 prompt_data["49"]["inputs"]["image"] = img[0] # 提示词 # prompt_data["131"]["inputs"]["String"] = prompt return prompt_data def C23(prompt_data, img): # 原图 prompt_data["52"]["inputs"]["image"] = img[0] # prompt_data["331"]["inputs"]["text"] = prompt return prompt_data def C24(prompt_data, img): # 原图 prompt_data["12"]["inputs"]["image"] = img[0] return prompt_data "source_images": [ "111,inputs,image" ], "style_pose_images": [ { "key": "113,inputs,image", "value": "./Image/C22/style_1.png" } ], def C25(prompt_data, img, model_id): # 原图 prompt_data["111"]["inputs"]["image"] = img[0] ####------------------------------------------------------------#### # 记录下用户的风格图+姿势图的组合，下次生成时，尽时避开已经输出过的组合，让用户更好的玩耍 # 获取目录下有多少个pose,多少个style directory_path = './Image/C' + str(model_id) + '/' pose_count = getPrefixFileCount(directory_path, 'pose_') style_count = getPrefixFileCount(directory_path, 'style_') style_id = random.randint(1, style_count) # 查找未使用过的姿态图与姿势图，如果找不到就随机用一个 pose_id = random.randint(1, max(pose_count, 1)) # for i in range(1, style_count + 1): # for j in range(1, pose_count + 1): # flag = checkStylePoseIdExists(user_id, model_id, i, j) # if not flag: # style_id = i # pose_id = j # # 回写数据库表，标识这些组合已经用过 # record_style_pose_id(user_id, model_id, style_id, pose_id) ####------------------------------------------------------------#### with open('./Image/C' + str(model_id) + '/style_' + str(style_id) + ".png", "rb") as f: y = upload_file(server_address, f, "", True) # 风格(随机) prompt_data["113"]["inputs"]["image"] = y return prompt_data # 获取C26中有哪些样式 def C26_get_sdxl_styles(): files = ['./JSON/C26/sdxl_styles_sai.json', './JSON/C26/sdxl_styles_twri.json'] SDXL_STYLES = [] for json_file in files: with open(json_file, 'r', encoding="utf-8") as fi: data = json.load(fi) for x in data: SDXL_STYLES.append(x['name']) return SDXL_STYLES def C26(prompt_data, img): # 原图 prompt_data["13"]["inputs"]["image"] = img[0] # 风格图片 SDXL_STYLES = C26_get_sdxl_styles() style_id = random.randint(0, len(SDXL_STYLES) - 1) pose_id = 1 # for i in range(0, len(SDXL_STYLES)): # if not checkStylePoseIdExists(user_id, model_id, i, 1): # style_id = i # # # 回写数据库表，标识这些组合已经用过 # record_style_pose_id(user_id, model_id, style_id, pose_id) prompt_data["79"]["inputs"]["style"] = SDXL_STYLES[style_id] return prompt_data def C27(prompt_data, img): # 原图 if len(img) == 1: with open('./Image/huanghai_3.png', "rb") as f: x = upload_file(server_address, f, "", True) prompt_data["82"]["inputs"]["image"] = x prompt_data["78"]["inputs"]["image"] = img[0] # 女 else: prompt_data["13"]["inputs"]["image"] = img[0] # 男 prompt_data["78"]["inputs"]["image"] = img[1] # 女 # 风格图片 style_id = random.randint(1, 2) pose_id = random.randint(1, 3) # for i in range(1, 3): # for j in range(1, 4): # if not checkStylePoseIdExists(user_id, model_id, i, j): # style_id = i # pose_id = j # # # 回写数据库表，标识这些组合已经用过 # record_style_pose_id(user_id, model_id, style_id, pose_id) # 姿势图有3个 pose_id pose_img_file = './Image/C27/pose_' + str(pose_id) + ".jpg" # 上传姿势图 with open(pose_img_file, "rb") as f: y = upload_file(server_address, f, "", True) prompt_data["82"]["inputs"]["image"] = y # 模型有2个 style_id if style_id == 1: prompt_data["4"]["inputs"]["ckpt_name"] = "samaritan3dCartoon_v40SDXL.safetensors" else: prompt_data["4"]["inputs"]["ckpt_name"] = "turbovisionxlSuperFastXLBasedOnNew_tvxlV431Bakedvae.safetensors" return prompt_data