如何从GIF中提取帧,并重建每一帧的细节?

3
我尝试从GIF文件中提取每一帧。
我发现有两种方法可以解决这个问题。

1 找一个在线工具来解决它。

https://ezgif.com/split

enter image description here

这是一个优秀的工具。它可以重新绘制细节。从GIF中提取的帧具有高质量。

enter image description here

2 尝试使用Python库来解决这个问题。

我使用了PIL,但它存在一个棘手的问题。提取的帧带有白边,丢失了许多细节。

enter image description here

我想问一下EZGif使用的算法是什么,以及如何用Python实现它?

请添加您用于提取帧的Python代码。您要提取哪种文件类型?我认为这与透明背景有关。尝试按此处所示保存为单个gif图像。链接 - agtoever
2个回答

3

我找到了一个有用的参考资料,与你遇到的问题相关。

import os
from PIL import Image


'''
I searched high and low for solutions to the "extract animated GIF frames in Python"
problem, and after much trial and error came up with the following solution based
on several partial examples around the web (mostly Stack Overflow).
There are two pitfalls that aren't often mentioned when dealing with animated GIFs -
firstly that some files feature per-frame local palettes while some have one global
palette for all frames, and secondly that some GIFs replace the entire image with
each new frame ('full' mode in the code below), and some only update a specific
region ('partial').
This code deals with both those cases by examining the palette and redraw
instructions of each frame. In the latter case this requires a preliminary (usually
partial) iteration of the frames before processing, since the redraw mode needs to
be consistently applied across all frames. I found a couple of examples of
partial-mode GIFs containing the occasional full-frame redraw, which would result
in bad renders of those frames if the mode assessment was only done on a
single-frame basis.
Nov 2012
'''


def analyseImage(path):
    '''
    Pre-process pass over the image to determine the mode (full or additive).
    Necessary as assessing single frames isn't reliable. Need to know the mode 
    before processing all frames.
    '''
    im = Image.open(path)
    results = {
        'size': im.size,
        'mode': 'full',
    }
    try:
        while True:
            if im.tile:
                tile = im.tile[0]
                update_region = tile[1]
                update_region_dimensions = update_region[2:]
                if update_region_dimensions != im.size:
                    results['mode'] = 'partial'
                    break
            im.seek(im.tell() + 1)
    except EOFError:
        pass
    return results


def processImage(path):
    '''
    Iterate the GIF, extracting each frame.
    '''
    mode = analyseImage(path)['mode']

    im = Image.open(path)

    i = 0
    p = im.getpalette()
    last_frame = im.convert('RGBA')

    try:
        while True:
            print "saving %s (%s) frame %d, %s %s" % (path, mode, i, im.size, im.tile)

            '''
            If the GIF uses local colour tables, each frame will have its own palette.
            If not, we need to apply the global palette to the new frame.
            '''
            if not im.getpalette():
                im.putpalette(p)

            new_frame = Image.new('RGBA', im.size)

            '''
            Is this file a "partial"-mode GIF where frames update a region of a different size to the entire image?
            If so, we need to construct the new frame by pasting it on top of the preceding frames.
            '''
            if mode == 'partial':
                new_frame.paste(last_frame)

            new_frame.paste(im, (0,0), im.convert('RGBA'))
            new_frame.save('%s-%d.png' % (''.join(os.path.basename(path).split('.')[:-1]), i), 'PNG')

            i += 1
            last_frame = new_frame
            im.seek(im.tell() + 1)
    except EOFError:
        pass


def main():
    processImage('foo.gif')
    processImage('bar.gif')


if __name__ == "__main__":
    main()

2

PIL中的GIF透明度存在问题。不确定为什么会这样,但这是事实。

你可以尝试使用我的GIF库,我刚刚为它创建了一个Python前端:

from PIL import Image

def GIF_Load(file):
    from platform import system
    from ctypes import string_at, Structure, c_long as cl, c_ubyte, \
                       py_object, pointer, POINTER as PT, CFUNCTYPE, CDLL
    class GIF_WHDR(Structure): _fields_ = \
       [("xdim", cl), ("ydim", cl), ("clrs", cl), ("bkgd", cl),
        ("tran", cl), ("intr", cl), ("mode", cl), ("frxd", cl), ("fryd", cl),
        ("frxo", cl), ("fryo", cl), ("time", cl), ("ifrm", cl), ("nfrm", cl),
        ("bptr", PT(c_ubyte)), ("cpal", PT(c_ubyte))]
    def intr(y, x, w, base, tran): base.paste(tran.crop((0, y, x, y + 1)), w)
    def skew(i, r): return r >> ((7 - (i & 2)) >> (1 + (i & 1)))
    def WriteFunc(d, w):
        cpal = string_at(w[0].cpal, w[0].clrs * 3)
        list = d.contents.value
        if (len(list) == 0):
            list.append(Image.new("RGBA", (w[0].xdim, w[0].ydim)))
        tail = len(list) - 1
        base = Image.frombytes("L", (w[0].frxd, w[0].fryd),
                               string_at(w[0].bptr, w[0].frxd * w[0].fryd))
        if (w[0].intr != 0):
            tran = base.copy()
            [intr(skew(y, y) + (skew(y, w[0].fryd - 1) + 1, 0)[(y & 7) == 0],
                  w[0].frxd, (0, y), base, tran) for y in range(w[0].fryd)]
        tran = Image.eval(base, lambda indx: (255, 0)[indx == w[0].tran])
        base.putpalette(cpal)
        list[tail].paste(base, (w[0].frxo, w[0].fryo), tran)
        list[tail].info = {"delay" : w[0].time}
        if (w[0].ifrm != (w[0].nfrm - 1)):
            list.append(list[max(0, tail - int(w[0].mode == 3))].copy())
            if (w[0].mode == 2):
                base = Image.new("L", (w[0].frxd, w[0].fryd), w[0].bkgd)
                base.putpalette(cpal)
                list[tail + 1].paste(base, (w[0].frxo, w[0].fryo))
    try: file = open(file, "rb")
    except IOError: return []
    file.seek(0, 2)
    size = file.tell()
    file.seek(0, 0)
    list = []
    CDLL(("%s.so", "%s.dll")[system() == "Windows"] % "gif_load"). \
    GIF_Load(file.read(), size,
             CFUNCTYPE(None, PT(py_object), PT(GIF_WHDR))(WriteFunc),
             None, pointer(py_object(list)), 0)
    file.close()
    return list

def GIF_Save(file, fext):
    list = GIF_Load("%s.gif" % file)
    [pict.save("%s_f%d.%s" % (file, indx, fext))
     for (indx, pict) in enumerate(list)]

GIF_Save("insert_gif_name_here_without_extension", "png")

网页内容由stack overflow 提供, 点击上面的
可以查看英文原文,
原文链接