Extract Images from Presentation Shapes in Python via Java

Overview

Images in a presentation can appear in several shape types: as ordinary picture frames, as picture fills applied to shapes, as OLE object preview images, as video or audio frame thumbnails, as zoom images, or as images nested inside table, chart, and SmartArt shapes. Aspose.Slides stores those images in the presentation image collection, exposed through ImageCollection and PPImage objects.

If you only need to export every image resource embedded in a presentation, iterate through Presentation.getImages. This article focuses on a different task: traversing shapes to find where images are used on slides, so the saved files can keep useful context such as the slide number, shape position, and source type (picture frame, fill image, media preview, OLE preview, or zoom image).

Shared Helper Functions

Save the shared helper functions below in image_helpers.py alongside the example scripts. They keep the examples short. save_original_image writes the original embedded bytes, chooses a safe extension from the MIME type, and skips duplicate image binaries by SHA-256 hash.

from pathlib import Path
import hashlib
import re

import jpype
import asposeslides

if not jpype.isJVMStarted():
    jpype.startJVM()

from asposeslides.api import FillType, GroupShape, ImageFormat


def save_original_image(image, output_directory, file_name_base, saved_image_hashes):
    image_data = bytes(image.getBinaryData())
    image_hash = hashlib.sha256(image_data).hexdigest()
    if image_hash in saved_image_hashes:
        return False
    saved_image_hashes.add(image_hash)
    extension = get_extension_from_content_type(image.getContentType())
    output_file = Path(output_directory) / f"{file_name_base}.{extension}"
    output_file.write_bytes(image_data)
    return True


def save_image_as_png(image, output_directory, file_name_base):
    output_file = Path(output_directory) / f"{file_name_base}.png"
    output_image = image.getImage()
    try:
        output_image.save(str(output_file), ImageFormat.Png)
    finally:
        output_image.dispose()


def get_picture_fill_image(fill_format):
    if fill_format is None or fill_format.getFillType() != FillType.Picture:
        return None
    return fill_format.getPictureFillFormat().getPicture().getImage()


def enumerate_shapes(shapes, prefix, include_grouped_shapes):
    shape_references = []
    for shape_index in range(shapes.size()):
        shape = shapes.get_Item(shape_index)
        shape_name_part = f"{prefix}_shape_{shape_index + 1}"
        shape_references.append((shape, shape_name_part))
        if include_grouped_shapes and isinstance(shape, GroupShape):
            child_shapes = shape.getShapes()
            child_references = enumerate_shapes(child_shapes, shape_name_part, include_grouped_shapes)
            shape_references.extend(child_references)
    return shape_references


def get_extension_from_content_type(content_type):
    if content_type is None or not str(content_type).strip():
        return "bin"
    media_type = str(content_type).split(";")[0].strip().lower()
    extensions = {
        "image/jpeg": "jpg",
        "image/png": "png",
        "image/gif": "gif",
        "image/bmp": "bmp",
        "image/tiff": "tiff",
        "image/x-emf": "emf",
        "image/emf": "emf",
        "image/x-wmf": "wmf",
        "image/wmf": "wmf",
        "image/svg+xml": "svg",
    }
    if media_type in extensions:
        return extensions[media_type]
    if media_type.startswith("image/"):
        return re.sub(r"[^A-Za-z0-9._-]", "_", media_type[len("image/"):])
    return "bin"

Extract Images from Picture Frames

Use this approach for pictures inserted as standalone objects. A PictureFrame provides access to its picture through getPictureFormat, getPicture, and getImage, which returns a PPImage object.

from pathlib import Path
import jpype
import asposeslides

if not jpype.isJVMStarted():
    jpype.startJVM()

from asposeslides.api import Presentation, PictureFrame
from image_helpers import enumerate_shapes, get_picture_fill_image, save_image_as_png, save_original_image

input_path = "sample.pptx"
output_directory = Path.cwd() / "extracted-images"
output_directory.mkdir(parents=True, exist_ok=True)
saved_image_hashes = set()

presentation = Presentation(input_path)
try:
    slide_count = presentation.getSlides().size()
    for slide_index in range(slide_count):
        slide = presentation.getSlides().get_Item(slide_index)
        slide_number = slide.getSlideNumber()
        slide_prefix = "slide_" + str(slide_number)
        shapes = slide.getShapes()
        shape_references = enumerate_shapes(shapes, slide_prefix, False)
        for shape, name_part in shape_references:
            if isinstance(shape, PictureFrame):
                picture_frame = shape
                image = picture_frame.getPictureFormat().getPicture().getImage()
                save_original_image(image, output_directory, name_part, saved_image_hashes)
finally:
    presentation.dispose()

Extract Images from Picture-Filled Shapes

Shapes can use a picture as their fill. Check the shape’s fill type first: if it is not FillType.Picture, there is no picture to extract from that fill. The example below handles AutoShape objects and saves each image as PNG through PPImage.getImage.

from pathlib import Path
import jpype
import asposeslides

if not jpype.isJVMStarted():
    jpype.startJVM()

from asposeslides.api import Presentation, AutoShape
from image_helpers import enumerate_shapes, get_picture_fill_image, save_image_as_png, save_original_image

input_path = "sample.pptx"
output_directory = Path.cwd() / "shape-fill-images"
output_directory.mkdir(parents=True, exist_ok=True)
saved_image_hashes = set()

presentation = Presentation(input_path)
try:
    slide_count = presentation.getSlides().size()
    for slide_index in range(slide_count):
        slide = presentation.getSlides().get_Item(slide_index)
        slide_number = slide.getSlideNumber()
        slide_prefix = "slide_" + str(slide_number)
        shapes = slide.getShapes()
        shape_references = enumerate_shapes(shapes, slide_prefix, False)
        for shape, name_part in shape_references:
            if isinstance(shape, AutoShape):
                auto_shape = shape
                fill_format = auto_shape.getFillFormat()
                image = get_picture_fill_image(fill_format)
                if image is not None:
                    save_image_as_png(image, output_directory, name_part)
finally:
    presentation.dispose()

Extract Preview Images from OLE Object Frames

An OleObjectFrame can have a substitute picture that PowerPoint uses as the object’s preview on a slide. This image is available through getSubstitutePictureFormat, getPicture, and getImage. Extracting this picture gives you the preview image, not the embedded OLE package contents.

from pathlib import Path
import jpype
import asposeslides

if not jpype.isJVMStarted():
    jpype.startJVM()

from asposeslides.api import Presentation, OleObjectFrame
from image_helpers import enumerate_shapes, get_picture_fill_image, save_image_as_png, save_original_image

input_path = "sample.pptx"
output_directory = Path.cwd() / "ole-preview-images"
output_directory.mkdir(parents=True, exist_ok=True)
saved_image_hashes = set()

presentation = Presentation(input_path)
try:
    slide_count = presentation.getSlides().size()
    for slide_index in range(slide_count):
        slide = presentation.getSlides().get_Item(slide_index)
        slide_number = slide.getSlideNumber()
        slide_prefix = "slide_" + str(slide_number)
        shapes = slide.getShapes()
        shape_references = enumerate_shapes(shapes, slide_prefix, False)
        for shape, name_part in shape_references:
            if isinstance(shape, OleObjectFrame):
                ole_object_frame = shape
                image = ole_object_frame.getSubstitutePictureFormat().getPicture().getImage()
                if image is not None:
                    file_name_base = name_part + "_ole_preview"
                    save_original_image(image, output_directory, file_name_base, saved_image_hashes)
finally:
    presentation.dispose()

Extract Preview Images from Video Frames

A VideoFrame can also store a preview image in getPictureFormat, getPicture, and getImage. This is the poster or thumbnail shown on the slide, not a frame decoded from the video stream.

from pathlib import Path
import jpype
import asposeslides

if not jpype.isJVMStarted():
    jpype.startJVM()

from asposeslides.api import Presentation, VideoFrame
from image_helpers import enumerate_shapes, get_picture_fill_image, save_image_as_png, save_original_image

input_path = "sample.pptx"
output_directory = Path.cwd() / "video-preview-images"
output_directory.mkdir(parents=True, exist_ok=True)
saved_image_hashes = set()

presentation = Presentation(input_path)
try:
    slide_count = presentation.getSlides().size()
    for slide_index in range(slide_count):
        slide = presentation.getSlides().get_Item(slide_index)
        slide_number = slide.getSlideNumber()
        slide_prefix = "slide_" + str(slide_number)
        shapes = slide.getShapes()
        shape_references = enumerate_shapes(shapes, slide_prefix, False)
        for shape, name_part in shape_references:
            if isinstance(shape, VideoFrame):
                video_frame = shape
                image = video_frame.getPictureFormat().getPicture().getImage()
                if image is not None:
                    file_name_base = name_part + "_video_preview"
                    save_original_image(image, output_directory, file_name_base, saved_image_hashes)
finally:
    presentation.dispose()

Extract Preview Images from Audio Frames

An AudioFrame can store a thumbnail in getPictureFormat, getPicture, and getImage. This is the image shown for the audio object on the slide.

from pathlib import Path
import jpype
import asposeslides

if not jpype.isJVMStarted():
    jpype.startJVM()

from asposeslides.api import Presentation, AudioFrame
from image_helpers import enumerate_shapes, get_picture_fill_image, save_image_as_png, save_original_image

input_path = "sample.pptx"
output_directory = Path.cwd() / "audio-preview-images"
output_directory.mkdir(parents=True, exist_ok=True)
saved_image_hashes = set()

presentation = Presentation(input_path)
try:
    slide_count = presentation.getSlides().size()
    for slide_index in range(slide_count):
        slide = presentation.getSlides().get_Item(slide_index)
        slide_number = slide.getSlideNumber()
        slide_prefix = "slide_" + str(slide_number)
        shapes = slide.getShapes()
        shape_references = enumerate_shapes(shapes, slide_prefix, False)
        for shape, name_part in shape_references:
            if isinstance(shape, AudioFrame):
                audio_frame = shape
                image = audio_frame.getPictureFormat().getPicture().getImage()
                if image is not None:
                    file_name_base = name_part + "_audio_preview"
                    save_original_image(image, output_directory, file_name_base, saved_image_hashes)
finally:
    presentation.dispose()

Extract Images from Zoom Objects

ZoomFrame and SectionZoomFrame shapes can use custom images. Read getZoomImage from the zoom frame.

from pathlib import Path
import jpype
import asposeslides

if not jpype.isJVMStarted():
    jpype.startJVM()

from asposeslides.api import Presentation, SectionZoomFrame, ZoomFrame
from image_helpers import enumerate_shapes, get_picture_fill_image, save_image_as_png, save_original_image

input_path = "sample.pptx"
output_directory = Path.cwd() / "zoom-images"
output_directory.mkdir(parents=True, exist_ok=True)
saved_image_hashes = set()

presentation = Presentation(input_path)
try:
    slide_count = presentation.getSlides().size()
    for slide_index in range(slide_count):
        slide = presentation.getSlides().get_Item(slide_index)
        slide_number = slide.getSlideNumber()
        slide_prefix = "slide_" + str(slide_number)
        shapes = slide.getShapes()
        shape_references = enumerate_shapes(shapes, slide_prefix, False)
        for shape, name_part in shape_references:
            if isinstance(shape, ZoomFrame):
                zoom_frame = shape
                image = zoom_frame.getZoomImage()
                if image is not None:
                    file_name_base = name_part + "_zoom"
                    save_original_image(image, output_directory, file_name_base, saved_image_hashes)
                    continue
            if isinstance(shape, SectionZoomFrame):
                section_zoom_frame = shape
                image = section_zoom_frame.getZoomImage()
                if image is not None:
                    file_name_base = name_part + "_section_zoom"
                    save_original_image(image, output_directory, file_name_base, saved_image_hashes)
                    continue
finally:
    presentation.dispose()

Extract Images from Summary Zoom Frames

A SummaryZoomFrame is also a shape. Its section items can use custom images, exposed through each summary zoom section’s getZoomImage method.

from pathlib import Path
import jpype
import asposeslides

if not jpype.isJVMStarted():
    jpype.startJVM()

from asposeslides.api import Presentation, SummaryZoomFrame
from image_helpers import enumerate_shapes, get_picture_fill_image, save_image_as_png, save_original_image

input_path = "sample.pptx"
output_directory = Path.cwd() / "summary-zoom-images"
output_directory.mkdir(parents=True, exist_ok=True)
saved_image_hashes = set()

presentation = Presentation(input_path)
try:
    slide_count = presentation.getSlides().size()
    for slide_index in range(slide_count):
        slide = presentation.getSlides().get_Item(slide_index)
        slide_number = slide.getSlideNumber()
        slide_prefix = "slide_" + str(slide_number)
        shapes = slide.getShapes()
        shape_references = enumerate_shapes(shapes, slide_prefix, False)
        for shape, name_part in shape_references:
            if isinstance(shape, SummaryZoomFrame):
                summary_zoom_frame = shape
                section_count = summary_zoom_frame.getSummaryZoomCollection().size()
                for section_index in range(section_count):
                    section = summary_zoom_frame.getSummaryZoomCollection().get_Item(section_index)
                    image = section.getZoomImage()
                    if image is not None:
                        display_index = section_index + 1
                        file_name_base = name_part + "_summary_zoom_" + str(display_index)
                        save_original_image(image, output_directory, file_name_base, saved_image_hashes)
finally:
    presentation.dispose()

Extract Images from Table Shapes

A Table is a shape. Images in a table are usually stored as picture fills in table cells.

from pathlib import Path
import jpype
import asposeslides

if not jpype.isJVMStarted():
    jpype.startJVM()

from asposeslides.api import Presentation, Table
from image_helpers import enumerate_shapes, get_picture_fill_image, save_image_as_png, save_original_image

input_path = "sample.pptx"
output_directory = Path.cwd() / "table-images"
output_directory.mkdir(parents=True, exist_ok=True)
saved_image_hashes = set()

presentation = Presentation(input_path)
try:
    slide_count = presentation.getSlides().size()
    for slide_index in range(slide_count):
        slide = presentation.getSlides().get_Item(slide_index)
        slide_number = slide.getSlideNumber()
        slide_prefix = "slide_" + str(slide_number)
        shapes = slide.getShapes()
        shape_references = enumerate_shapes(shapes, slide_prefix, True)
        for shape, name_part in shape_references:
            if isinstance(shape, Table):
                table = shape
                row_count = table.getRows().size()
                column_count = table.getColumns().size()
                for row_index in range(row_count):
                    for column_index in range(column_count):
                        cell = table.get_Item(column_index, row_index)
                        fill_format = cell.getCellFormat().getFillFormat()
                        image = get_picture_fill_image(fill_format)
                        if image is not None:
                            display_row = row_index + 1
                            display_column = column_index + 1
                            file_name_base = name_part + "_cell_" + str(display_row) + "_" + str(display_column)
                            save_original_image(image, output_directory, file_name_base, saved_image_hashes)
finally:
    presentation.dispose()

Extract Images from Chart Shapes

A Chart is a shape. The example below extracts an image from the chart area’s picture fill.

from pathlib import Path
import jpype
import asposeslides

if not jpype.isJVMStarted():
    jpype.startJVM()

from asposeslides.api import Presentation, Chart
from image_helpers import enumerate_shapes, get_picture_fill_image, save_image_as_png, save_original_image

input_path = "sample.pptx"
output_directory = Path.cwd() / "chart-images"
output_directory.mkdir(parents=True, exist_ok=True)
saved_image_hashes = set()

presentation = Presentation(input_path)
try:
    slide_count = presentation.getSlides().size()
    for slide_index in range(slide_count):
        slide = presentation.getSlides().get_Item(slide_index)
        slide_number = slide.getSlideNumber()
        slide_prefix = "slide_" + str(slide_number)
        shapes = slide.getShapes()
        shape_references = enumerate_shapes(shapes, slide_prefix, True)
        for shape, name_part in shape_references:
            if isinstance(shape, Chart):
                chart = shape
                fill_format = chart.getFillFormat()
                image = get_picture_fill_image(fill_format)
                if image is not None:
                    file_name_base = name_part + "_chart_area"
                    save_original_image(image, output_directory, file_name_base, saved_image_hashes)
finally:
    presentation.dispose()

Extract Images from SmartArt Shapes

A SmartArt object is a shape. Depending on the SmartArt layout, images may be stored in node bullet fills or in the fill formats of node shapes.

from pathlib import Path
import jpype
import asposeslides

if not jpype.isJVMStarted():
    jpype.startJVM()

from asposeslides.api import Presentation, SmartArt
from image_helpers import enumerate_shapes, get_picture_fill_image, save_image_as_png, save_original_image

input_path = "sample.pptx"
output_directory = Path.cwd() / "smartart-images"
output_directory.mkdir(parents=True, exist_ok=True)
saved_image_hashes = set()

presentation = Presentation(input_path)
try:
    slide_count = presentation.getSlides().size()
    for slide_index in range(slide_count):
        slide = presentation.getSlides().get_Item(slide_index)
        slide_number = slide.getSlideNumber()
        slide_prefix = "slide_" + str(slide_number)
        shapes = slide.getShapes()
        shape_references = enumerate_shapes(shapes, slide_prefix, True)
        for shape, name_part in shape_references:
            if isinstance(shape, SmartArt):
                smart_art = shape
                node_count = smart_art.getAllNodes().size()
                for node_index in range(node_count):
                    node = smart_art.getAllNodes().get_Item(node_index)
                    bullet_fill_format = node.getBulletFillFormat()
                    bullet_image = get_picture_fill_image(bullet_fill_format)
                    if bullet_image is not None:
                        display_node = node_index + 1
                        file_name_base = name_part + "_smartart_node_" + str(display_node) + "_bullet"
                        save_original_image(bullet_image, output_directory, file_name_base, saved_image_hashes)
                    node_shape_count = node.getShapes().size()
                    for node_shape_index in range(node_shape_count):
                        node_shape = node.getShapes().get_Item(node_shape_index)
                        fill_format = node_shape.getFillFormat()
                        image = get_picture_fill_image(fill_format)
                        if image is not None:
                            display_node = node_index + 1
                            display_node_shape = node_shape_index + 1
                            file_name_base = name_part + "_smartart_node_" + str(display_node) + "_shape_" + str(display_node_shape)
                            save_original_image(image, output_directory, file_name_base, saved_image_hashes)
finally:
    presentation.dispose()

Include Images Inside Grouped Shapes

Grouped shapes contain their own shape collections. The shared enumerate_shapes helper has an include_grouped_shapes option. Set it to True when you want to inspect shapes inside GroupShape objects. The example below extracts images from picture frames, picture-filled shapes, OLE object previews, video frame thumbnails, and audio frame thumbnails. To include table, chart, SmartArt, and summary zoom images as well, reuse the specialized extraction logic from the previous sections while keeping the same recursive shape traversal.

from pathlib import Path
import jpype
import asposeslides

if not jpype.isJVMStarted():
    jpype.startJVM()

from asposeslides.api import Presentation, AudioFrame, AutoShape, OleObjectFrame, PictureFrame, VideoFrame
from image_helpers import enumerate_shapes, get_picture_fill_image, save_image_as_png, save_original_image

input_path = "sample.pptx"
output_directory = Path.cwd() / "all-shape-images"
output_directory.mkdir(parents=True, exist_ok=True)
saved_image_hashes = set()

presentation = Presentation(input_path)
try:
    slide_count = presentation.getSlides().size()
    for slide_index in range(slide_count):
        slide = presentation.getSlides().get_Item(slide_index)
        slide_number = slide.getSlideNumber()
        slide_prefix = "slide_" + str(slide_number)
        shapes = slide.getShapes()
        shape_references = enumerate_shapes(shapes, slide_prefix, True)
        for shape, name_part in shape_references:
            if isinstance(shape, OleObjectFrame):
                ole_object_frame = shape
                image = ole_object_frame.getSubstitutePictureFormat().getPicture().getImage()
                if image is not None:
                    file_name_base = name_part + "_ole_preview"
                    save_original_image(image, output_directory, file_name_base, saved_image_hashes)
                continue
            if isinstance(shape, VideoFrame):
                video_frame = shape
                image = video_frame.getPictureFormat().getPicture().getImage()
                if image is not None:
                    file_name_base = name_part + "_video_preview"
                    save_original_image(image, output_directory, file_name_base, saved_image_hashes)
                continue
            if isinstance(shape, AudioFrame):
                audio_frame = shape
                image = audio_frame.getPictureFormat().getPicture().getImage()
                if image is not None:
                    file_name_base = name_part + "_audio_preview"
                    save_original_image(image, output_directory, file_name_base, saved_image_hashes)
                continue
            if isinstance(shape, PictureFrame):
                picture_frame = shape
                image = picture_frame.getPictureFormat().getPicture().getImage()
                save_original_image(image, output_directory, name_part, saved_image_hashes)
                continue
            if isinstance(shape, AutoShape):
                auto_shape = shape
                fill_format = auto_shape.getFillFormat()
                image = get_picture_fill_image(fill_format)
                if image is not None:
                    save_original_image(image, output_directory, name_part, saved_image_hashes)
finally:
    presentation.dispose()

Edge Cases and Practical Notes

  • Duplicate images: Multiple shapes may reference the same image or separate images with identical bytes. Hash PPImage.getBinaryData before writing files if you want one output file per unique image.
  • Original data vs. converted output: Saving PPImage.getBinaryData preserves the embedded JPEG, PNG, GIF, SVG, EMF, or WMF data. Saving PPImage.getImage through save is useful when you want a consistent output format.
  • Unsupported fill types: Solid, gradient, pattern, and no-fill shapes do not contain a picture fill. Check FillType before reading getPictureFillFormat.
  • Grouped shapes: The top-level slide shape collection does not flatten groups. Recursively inspect GroupShape.getShapes when grouped content matters.
  • OLE object previews: An OleObjectFrame may expose a preview image through getSubstitutePictureFormat, but that image is only the slide preview. It is not the embedded file inside the OLE object.
  • Video frame thumbnails: A VideoFrame may expose a preview image through getPictureFormat, but that image is only the poster shown on the slide. It is not extracted from the video stream.
  • Audio frame thumbnails: An AudioFrame may expose an icon or thumbnail through getPictureFormat; it is not the embedded audio data.
  • Zoom images: Slide zoom, section zoom, and summary zoom shapes may use custom PPImage objects through getZoomImage.
  • Nested shape models: Table, chart, and SmartArt objects implement Shape, but their images are often stored in nested table cell, chart element, or SmartArt node formatting objects.
  • Cropped or transformed pictures: Accessing PPImage gives you the stored image resource. It does not render cropping, transparency, recoloring, rotation, or other visual effects applied by the shape.

FAQ

Can I extract the original image without cropping, effects, or shape transformations?

Yes. Access the PPImage object and write PPImage.getBinaryData to disk. This preserves the original encoded image stored in the presentation, not the way the image is rendered on the slide.

Can I export every extracted image as PNG?

Yes. Use PPImage.getImage to get an image object, and then call save with ImageFormat.Png. This converts the output and may not preserve the original file type or vector data.

How do I avoid saving the same image more than once?

Use a hash of PPImage.getBinaryData and keep the hashes in a set. If a new image has a hash that already exists, skip it or record another reference to the existing output file.

Why do some shapes not produce an image?

Picture frames, picture-filled shapes, OLE object frames, media frames, zoom frames, tables, charts, and SmartArt objects can reference images. Some shape types expose images through nested formatting objects, so a simple getPictureFormat or shape getFillFormat check is not always enough.

Can I extract the thumbnail shown for a video frame?

Yes. Use VideoFrame and read getPictureFormat, getPicture, and getImage. This extracts the poster image stored with the video frame, not a frame generated from the video file.

How can I determine which shapes use a specific image from the presentation image collection?

Aspose.Slides does not store reverse links from PPImage to shapes. Build a mapping during traversal: whenever you find an image reference, record the slide number, shape path, and image hash or collection item.

Can I extract images embedded inside OLE objects, such as attached documents?

You can extract the OLE object’s slide preview from OleObjectFrame.getSubstitutePictureFormat. However, that preview is not the embedded document itself. To extract images from inside the embedded file, extract the OLE data and inspect it with tools for that file type.

Get the Entire Slide Background from a Presentation as an Image Resize Shapes on Presentation Slides in Python via Java