Now it only output the final img with ‘yes’ on it.
But i do want to add all pictures into one single image.
This is a black-white image, generated with font.
enter image description here
prompt: text_testnyes, it’s just for test

texts = str(prompt).split('n')
        n_lines = len(texts)
        if len(texts ) == 0:
            texts  = [' ']
        max_chars = 20
        font_path = os.path.join(folder_paths.models_dir, "fonts", font)
        self.font = ImageFont.truetype(font_path, size=60, encoding='utf-8')
        mask_img = tensor2numpy_cv2(pos_mask)
        mask_img = cv2.cvtColor(mask_img, cv2.COLOR_GRAY2RGB) # cv2二值图(mask)转rgb
        mask_img = cv2.bitwise_not(mask_img) # cv2图片取反
        
        from .AnyText_scripts.AnyText_pipeline_util import resize_image
        from .AnyText_scripts.AnyText_t3_dataset import draw_glyph2
        from .AnyText_scripts.AnyText_pipeline import separate_pos_imgs, find_polygon
        draw_pos = resize_image(mask_img, max_length=768)
        draw_pos = cv2.resize(draw_pos, (width, height))
        pos_imgs = 255-draw_pos
        pos_imgs = pos_imgs[..., 0:1]
        pos_imgs = cv2.convertScaleAbs(pos_imgs)
        _, pos_imgs = cv2.threshold(pos_imgs, 254, 255, cv2.THRESH_BINARY)
        pos_imgs = separate_pos_imgs(pos_imgs, sort_radio)
        if len(pos_imgs) == 0:
            pos_imgs = [np.zeros((height, width, 1))]
        if len(pos_imgs) < n_lines:
            if n_lines == 1 and texts[0] == ' ':
                # pass  # text-to-image without text
                print('33[93m', f'Warning: text-to-image without text.', '33[0m')
            else:
                # return None, -1, f'Found {len(pos_imgs)} positions that < needed {n_lines} from prompt, check and try again!', ''
                raise ValueError(f'Found {len(pos_imgs)} positions that < needed {n_lines} from prompt, check and try again(手绘遮罩数少于要绘制的文本数，检查再重试)!')
        elif len(pos_imgs) > n_lines:
            print('33[93m', f'Warning: found {len(pos_imgs)} positions that > needed {n_lines} from prompt.', '33[0m')
        pre_pos = []
        poly_list = []
        for input_pos in pos_imgs:
            if input_pos.mean() != 0:
                input_pos = input_pos[..., np.newaxis] if len(input_pos.shape) == 2 else input_pos
                poly, pos_img = find_polygon(input_pos)
                pre_pos += [pos_img/255.]
                poly_list += [poly]
            else:
                pre_pos += [np.zeros((height, width, 1))]
                poly_list += [None]
        for i in range(len(texts)):
            text = texts[i]
            if len(text) > max_chars:
                text = text[:max_chars]
            gly_scale = 2
            if pre_pos[i].mean() != 0:
                _, glyphs = draw_glyph2(self.font, text, poly_list[i], scale=gly_scale, width=width, height=height, add_space=False)
        font_img = pil2tensor(glyphs)
        
        return (font_img, )

def draw_glyph2(font, text, polygon, vertAng=10, scale=1, width=512, height=512, add_space=True):
    enlarge_polygon = polygon*scale
    rect = cv2.minAreaRect(enlarge_polygon)
    box = cv2.boxPoints(rect)
    box = np.int0(box)
    w, h = rect[1]
    angle = rect[2]
    if angle < -45:
        angle += 90
    angle = -angle
    if w < h:
        angle += 90

    vert = False
    if (abs(angle) % 90 < vertAng or abs(90-abs(angle) % 90) % 90 < vertAng):
        _w = max(box[:, 0]) - min(box[:, 0])
        _h = max(box[:, 1]) - min(box[:, 1])
        if _h >= _w:
            vert = True
            angle = 0

    img = np.zeros((height*scale, width*scale, 3), np.uint8)
    img = Image.fromarray(img)

    # infer font size
    image4ratio = Image.new("RGB", img.size, "white")
    draw = ImageDraw.Draw(image4ratio)
    _, _, _tw, _th = draw.textbbox(xy=(0, 0), text=text, font=font)
    text_w = min(w, h) * (_tw / _th)
    if text_w <= max(w, h):
        # add space
        if len(text) > 1 and not vert and add_space:
            for i in range(1, 100):
                text_space = insert_spaces(text, i)
                _, _, _tw2, _th2 = draw.textbbox(xy=(0, 0), text=text_space, font=font)
                if min(w, h) * (_tw2 / _th2) > max(w, h):
                    break
            text = insert_spaces(text, i-1)
        font_size = min(w, h)*0.80
    else:
        shrink = 0.75 if vert else 0.85
        font_size = min(w, h) / (text_w/max(w, h)) * shrink
    # print('33[93m', font_size, 'n', type(font_size), '33[0m')
    if not font_size > 0:
        font_size = min(max(1, int(font_size)), 100) #可以防止报错：font size must be greater than 0
    new_font = font.font_variant(size=int(font_size))

    left, top, right, bottom = new_font.getbbox(text)
    text_width = right-left
    text_height = bottom - top

    layer = Image.new('RGBA', img.size, (0, 0, 0, 0))
    draw = ImageDraw.Draw(layer)
    if not vert:
        draw.text((rect[0][0]-text_width//2, rect[0][1]-text_height//2-top), text, font=new_font, fill=(255, 255, 255, 255))
    else:
        x_s = min(box[:, 0]) + _w//2 - text_height//2
        y_s = min(box[:, 1])
        for c in text:
            draw.text((x_s, y_s), c, font=new_font, fill=(255, 255, 255, 255))
            _, _t, _, _b = new_font.getbbox(c)
            y_s += _b

    rotated_layer = layer.rotate(angle, expand=1, center=(rect[0][0], rect[0][1]))

    x_offset = int((img.width - rotated_layer.width) / 2)
    y_offset = int((img.height - rotated_layer.height) / 2)
    img.paste(rotated_layer, (x_offset, y_offset), rotated_layer)
    cv2_img = img
    img = np.expand_dims(np.array(img.convert('1')), axis=2).astype(np.float64)
    return (img, cv2_img, )

Tried for loop but i’m not good at it, failed.
Just one step for all text in one image.