Add new TextClip with original text shape

[GrayFL]

While using version 2.x of moviepy, I noticed that the position of TextClip has changed in all videos generated using past code;

Analyzing it, I found that TextClip now uses PIL's multiline_textbbox to look for borders and weed out the excess. But then 'LINE' and 'xxxx' will not have the same shape (I mean '■■■■'), and I can't control two different pieces of text with different content to be in the same position just by using with_position

So I'd like to be able to provide the raw image output by PIL without any complicated margin and text_size calculations, like the following:

class SimpleTextClip(ImageClip):
    @convert_path_to_string("filename")
    def __init__(
        self,
        font,
        text=None,
        filename=None,
        font_size=None,
        # size=(None, None),
        # margin=(None, None),
        color="black",
        bg_color=None,
        stroke_color=None,
        stroke_width=0,
        method="label",
        text_align="left",
        # horizontal_align="center",
        # vertical_align="center",
        interline=4,
        transparent=True,
        duration=None,
    ):

        try:
            _ = ImageFont.truetype(font)
        except Exception as e:
            raise ValueError(
                "Invalid font {}, pillow failed to use it with error {}".format(font, e)
            )

        if filename:
            with open(filename, "r") as file:
                text = file.read().rstrip()  # Remove newline at end

        if text is None:
            raise ValueError("No text nor filename provided")

        # Trace the image
        img_mode = "RGBA" if transparent else "RGB"

        if bg_color is None and transparent:
            bg_color = (0, 0, 0, 0)

        pil_font = ImageFont.truetype(font, font_size)

        left, top, right, bottom = ImageDraw.Draw(Image.new("RGB", (1, 1))).multiline_textbbox(
            (0,0),text=text,
            font=pil_font,
            spacing=interline,
            align=text_align,
            stroke_width=stroke_width,
            anchor="la",
            )
        img_width, img_height = int(np.ceil(right)), int((np.ceil(bottom)))
        # get the original shape of text
        img = Image.new(img_mode, (img_width, img_height), color=bg_color)
        draw = ImageDraw.Draw(img)

        draw.multiline_text(
            xy=(0, 0),
            text=text,
            fill=color,
            font=pil_font,
            spacing=interline,
            align=text_align,
            stroke_width=stroke_width,
            stroke_fill=stroke_color,
            anchor="la",
        )

        img_numpy = np.array(img)

        ImageClip.__init__(
            self, img=img_numpy, transparent=transparent, duration=duration
        )
        self.text = text
        self.color = color
        self.stroke_color = stroke_color

[RaSan147]

Also I am getting the text cut when generating by TextClip

Same code with v1.X

[harsh-zymr]

Facing the same issue. There are only 2 choices, either get the text cut like shown by @RaSan147 or have them up/down scattered.

Not sure how to align them well. In an earlier version, it was aligned properly. Maybe @GrayFL 's solution is the key. It does work.

[OsaAjani]

Hi, I absolutely hate the way textclip currently work, in my own defense text with Pillow is a pain second only to hitting your pinky in the table foot. I will try to improve things in near future, I hope I can achieve a more coherent behaviour and fix the current text cutting problem.

[OsaAjani]

Hi, making progress on that one, still some tests to run, but I think I should be able to provide a fix for the current text cutting/alignment problem in the next days.

[OsaAjani]

This issue is closely related to #2268. We're merging the discussion into that issue to keep everything organized. Please follow the updates there.