这些对图像文件的访问可能来自django.db.models.fields.files.ImageField.update_dimension_fields。
它的描述是这样的：

"""
Updates field's width and height fields, if defined.

This method is hooked up to model's post_init signal to update
dimensions after instantiating a model instance.  However, dimensions
won't be updated if the dimensions fields are already populated.  This
avoids unnecessary recalculation when loading an object from the
database.

Dimensions can be forced to update with force=True, which is how
ImageFileDescriptor.__set__ calls this method.
"""

字符串
在你的情况下，我会做的是扮演我们自己的width_field和height_field，并在图像上传时只更新一次。下面是我如何实现它（尚未在S3的生产环境中测试，但可以使用本地存储）：

class ImageCovered(models.Model):
    def __init__(self, *args, **kwargs):
        # read and store original image data
        super(ImageCovered, self).__init__(*args, **kwargs)
        self.__original_cover_image = self.cover_image

    cover_image = models.ImageField(upload_to=get_upload_path, max_length=200, null=True, blank=True,)
    # We should not use ImageField's width_field nor height_field here, since it might call S3 files.
    # Instead, we implement our own fields and update it once only upon image saving.
    cover_image_height = models.PositiveSmallIntegerField(default=0)
    cover_image_width = models.PositiveSmallIntegerField(default=0)

    def save(self):
        super(ImageCovered, self).save()
        # Check if cover_image has been changed
        if self.__original_cover_image != self.cover_image :
            self.cover_image_height = self.cover_image.height
            self.cover_image_width = self.cover_image.width
            self.__original_cover_image = self.cover_image #replace original data
            super(ImageCovered, self).save(update_fields=('cover_image_height', 'cover_image_width',))

型

唯一的变化，我做的是增加了width_field和height_field的ImageField保存的高度和宽度，所以我没有检索图像从S3得到这个信息。
晚了10年，但对于在此更改之前上传的任何文件，您可能需要在所有对象上调用相关模型的保存（）方法来填充字段。