ids_peak_ipl.Image in the current Python bindings never had a lowercase-snake-case constructor called create_from(...), so the import line

python
ipl_image = ids_peak_ipl.Image.create_from(buffer)

raises the AttributeError.
There are two supported ways to turn a finished Buffer into an Image:

---

1 · The recommended helper (ids_peak_ipl_extension.BufferToImage)

IDS ship a tiny helper module that does exactly the C++ peak::BufferTo<Image> conversion for you:

python
from ids_peak import ids_peak_ipl_extension  # NEW import

buffer  = datastream.WaitForFinishedBuffer(5_000)
ipl_img = ids_peak_ipl_extension.BufferToImage(buffer)  # shallow copy

BufferToImage is part of the official docs and keeps the call short.(en.ids-imaging.com)

---

2 · Explicit static factories on Image

If you really need a deep-copy or want to build an image manually, the binding exposes Pascal-Case factories that mirror the C++ names, e.g.

python
ipl_img = ids_peak_ipl.Image.CreateFromSizeAndBuffer(
    buffer.PixelFormat(),          # pixel format object
    buffer.BasePtr(),              # raw pointer
    buffer.Size(),                 # size in bytes
    buffer.Width(),
    buffer.Height()
)

Notice the uppercase C in CreateFrom….(PyPI)

---

Updating the rest of your capture loop

python
from ids_peak import ids_peak_ipl, ids_peak_ipl_extension
import numpy as np

ipl_converter = ids_peak_ipl.ImageConverter()

buffer  = datastream.WaitForFinishedBuffer(5_000)
ipl_img = ids_peak_ipl_extension.BufferToImage(buffer)

# convert to 4-channel BGRa8 (keep alpha so width*height*4 is easy to reshape)
converted = ipl_converter.Convert(
    ipl_img, ids_peak_ipl.PixelFormatName_BGRa8
)

h, w  = converted.Height(), converted.Width()
c     = converted.PixelFormat().NumChannels()   # =4 for BGRa8 ([en.ids-imaging.com](https://en.ids-imaging.com/manuals/ids-peak/ids-peak-ipl-documentation/2.14.0/en/classpeak_1_1ipl_1_1_pixel_format.html))
img_np = np.frombuffer(converted.Data(), dtype=np.uint8).reshape(h, w, c)

• get_numpy_1d_view() still works, but np.frombuffer(converted.Data(), …) avoids an extra copy and lets you choose the dtype.
• PixelFormat().NumChannels() is the portable way to find the channel count; PixelPerChannel never existed in the public API.

Queue the buffer back only after you finish using ipl_img/converted to avoid dangling memory.

That change alone should make your script run without throwing the AttributeError and start printing the ten frame sizes as intended.

以下提供一份已修正、可直接執行的完整範例程式碼。
主要變動：

1. 使用官方輔助工具 ids_peak_ipl_extension.BufferToImage() 將 Buffer 轉成 Image，取代不存在的 Image.create_from()。
2. 透過 np.frombuffer() 取得 NumPy 陣列，並用 PixelFormat().NumChannels() 來取得通道數。
3. 確保在完全處理完影像之後，才把 buffer 重新排入佇列 (QueueBuffer)，避免記憶體仍被占用。

python
# -*- coding: utf-8 -*-
"""
使用 IDS peak SDK 擷取 10 張影像，並轉成 NumPy 陣列顯示尺寸
"""

import sys
import numpy as np
from ids_peak import ids_peak, ids_peak_ipl, ids_peak_ipl_extension
# 若需要顯示影像，可安裝 opencv-python 並取消下列註解
# import cv2


def main():
    device = None
    datastream = None
    nodemap_remote_device = None

    # ---------- 初始化 IDS peak Library ----------
    try:
        ids_peak.Library.Initialize()
        print("IDS peak Library 初始化成功。")
    except Exception as e:
        print(f"初始化 IDS peak Library 失敗: {e}")
        return -1

    try:
        # ---------- 搜尋並開啟第一台相機 ----------
        device_manager = ids_peak.DeviceManager.Instance()
        device_manager.Update()

        if device_manager.Devices().empty():
            print("錯誤：找不到任何相機。請檢查相機連接與驅動程式。")
            return -1

        device = device_manager.Devices()[0].OpenDevice(
            ids_peak.DeviceAccessType_Control
        )
        print(f"成功開啟相機: {device.ModelName()}")

        # ---------- 建立並啟動資料流 ----------
        datastreams = device.DataStreams()
        if not datastreams:
            print("錯誤：相機不支援資料流。")
            return -1

        datastream = datastreams[0].OpenDataStream()
        print("成功開啟資料流。")

        nodemap_remote_device = device.RemoteDevice().NodeMaps()[0]

        payload_size = nodemap_remote_device.FindNode("PayloadSize").Value()
        buffer_count_max = datastream.NumBuffersAnnouncedMinRequired()
        for _ in range(buffer_count_max):
            buf = datastream.AllocAndAnnounceBuffer(payload_size)
            datastream.QueueBuffer(buf)

        datastream.StartAcquisition()
        print("資料流開始擷取...")

        nodemap_remote_device.FindNode("AcquisitionStart").Execute()
        nodemap_remote_device.FindNode("AcquisitionStart").WaitUntilDone()
        print("相機開始傳送影像...")

        # ---------- 建立影像轉換器 ----------
        ipl_converter = ids_peak_ipl.ImageConverter()

        print("\n--- 開始擷取 10 張影像 ---")
        for i in range(10):
            # 1. 等待單張完整 Buffer
            buffer = datastream.WaitForFinishedBuffer(5_000)

            # 2. 轉成 IDS peak IPL Image  (shallow copy，不複製畫素資料)
            ipl_image = ids_peak_ipl_extension.BufferToImage(buffer)

            # 3. 轉成 BGRa8，方便後續顯示或儲存
            converted = ipl_converter.Convert(
                ipl_image, ids_peak_ipl.PixelFormatName_BGRa8
            )

            # 4. 取得 NumPy 陣列
            h, w = converted.Height(), converted.Width()
            channels = converted.PixelFormat().NumChannels()
            image_np = np.frombuffer(
                converted.Data(), dtype=np.uint8
            ).reshape(h, w, channels)

            print(
                f"成功擷取第 {i + 1}/10 張影像，尺寸 (HxW): "
                f"{image_np.shape[0]}x{image_np.shape[1]}"
            )

            # 需要顯示時可用下列程式
            # cv2.imshow("Frame", image_np)
            # cv2.waitKey(1)

            # 5. 使用完畢後再將 Buffer 排回佇列
            datastream.QueueBuffer(buffer)

        print("--- 影像擷取完成 ---\n")

    except Exception as e:
        print(f"程式執行期間發生錯誤: {e}")
        return -1

    finally:
        # ---------- 清理與釋放資源 ----------
        print("開始清理與釋放資源...")
        if device is not None and nodemap_remote_device is not None:
            try:
                nodemap_remote_device.FindNode("AcquisitionStop").Execute()
            except Exception:
                pass

        if datastream is not None:
            try:
                datastream.StopAcquisition(ids_peak.AcquisitionStopMode_Default)
                datastream.Flush(ids_peak.DataStreamFlushMode_DiscardAll)
                for buf in datastream.AnnouncedBuffers():
                    datastream.RevokeBuffer(buf)
            except Exception:
                pass

        try:
            ids_peak.Library.Close()
            print("IDS peak Library 已成功關閉。")
        except Exception as e:
            print(f"關閉 IDS peak Library 時發生錯誤: {e}")

        print("程式執行完畢。")

    return 0


if __name__ == "__main__":
    sys.exit(main())

使用說明

• 如需顯示畫面，請安裝 opencv-python 並取消 cv2 相關註解。
• 若想改存成 PNG/JPEG，或用其他像素格式，只需調整 Convert() 的目標格式與後續處理即可。

執行後，終端機應會列印十張影像的解析度，並再正常結束不拋出 AttributeError。祝開發順利!