JPEGE Functions and Restrictions

The Atlas training product does not support the JPEGE function described in this section.

Description

JPEG Encoder (JPEGE) encodes YUV images into JPEG images, for example, .jpg images.

Resolution Restrictions

Input image resolution range

Version	Resolution Range
Atlas inference product Atlas A2 training product / Atlas A2 inference product Atlas A3 training product / Atlas A3 inference product	32 x 32 to 8192 x 8192
Atlas 200I/500 A2 inference product	32 x 32 to 16384 x 16384

Version

Resolution Range

Atlas inference product

Atlas A2 training product / Atlas A2 inference product

Atlas A3 training product / Atlas A3 inference product

32 x 32 to 8192 x 8192

Atlas 200I/500 A2 inference product

32 x 32 to 16384 x 16384

Output resolution range:
JPEGE only encodes images and does not change the image resolution. Therefore, the resolution of the output image is the same as that of the input image.

Restrictions on Buffer Allocation and Freeing APIs

The size of the input buffer is the size of the input stream. For details about the size of the output buffer, see the formula in Table 1.

Version	Input Buffer Allocation/Freeing API	Input Buffer Allocation/Freeing API
Atlas inference product	During JPEGE image encoding, call acl.himpi.dvpp_malloc/acl.himpi.dvpp_free to allocate or free the input buffer on the device. The buffer lifetime is managed by the user.	The following two methods are supported: Managed by DVPP: Call acl.himpi.venc_send_frame to send source images for encoding. Correctly set the attribute value of hi_venc_chn_attr["venc_attr"]["buf_size"] When creating a channel by calling acl.himpi.venc_create_chn. For the parameter description, see hi_venc_attr. In this mode, the JPEG header of the output data does not contain the COM comment field, and the data length is smaller. However, you need to copy the output data from the buffer returned by DVPP to the specified memory. Managed by users: Call acl.himpi.venc_send_jpege_frame to send source images for encoding. When creating a channel by calling acl.himpi.venc_create_chn, set hi_venc_chn_attr["venc_attr"]["buf_size"] to 0 (for details about the parameter description, see hi_venc_attr). Then call acl.himpi.venc_get_jpege_predicted_size to estimate the output memory size, and call acl.himpi.dvpp_malloc or acl.himpi.dvpp_free to allocate or free the output memory. In this mode, the output buffer address is set when acl.himpi.venc_send_jpege_frame is called. The output data is directly stored in the buffer set by the user. Compared with the other mode, this mode does not require you to copy the output data from the buffer returned by DVPP to the specified memory. However, the JPEG header of the output data may contain the COM comment field (whose length ranges from 4 bytes to 19 bytes), and thus the data length is larger.
Atlas A2 training product / Atlas A2 inference product Atlas 200I/500 A2 inference product Atlas A3 training product / Atlas A3 inference product	During JPEGE image encoding, the following two types of buffer allocation/freeing APIs are supported: Allocate or free the input buffer on the device by calling acl.rt.malloc/acl.rt.free. The buffer lifetime is managed by the user. Allocate or free the input buffer on the device by calling acl.himpi.dvpp_malloc/acl.himpi.dvpp_free. The buffer lifetime is managed by the user. Note: The memory allocated by calling the acl.himpi.dvpp_malloc API is dedicated for media data processing. However, the address space of the dedicated memory is limited. If memory planning is concerned or memory resources are limited, you are advised to call the acl.rt.malloc API to allocate memory.	The following two methods are supported: Managed by DVPP: Call acl.himpi.venc_send_frame to send source images for encoding. Correctly set the attribute value of hi_venc_chn_attr["venc_attr"]["buf_size"] When creating a channel by calling acl.himpi.venc_create_chn. For the parameter description, see hi_venc_attr. In this mode, the JPEG header of the output data does not contain the COM comment field, and the data length is smaller. However, you need to copy the output data from the buffer returned by DVPP to the specified memory. Managed by users: Call acl.himpi.venc_send_jpege_frame to send source images for encoding. When calling the acl.himpi.venc_create_chn API to create a channel, set hi_venc_chn_attr["venc_attr"]["buf_size"] to 0 (for details about the parameter description, see hi_venc_attr). Then call the acl.himpi.venc_get_jpege_predicted_size API to estimate the output buffer size, and call the acl.himpi.dvpp_malloc\acl.himpi.dvpp_free API to allocate or free the output buffer. (You can also call the acl.rt.malloc\acl.rt.free API to allocate or free the output buffer on the device.) In this mode, the output buffer address is set when acl.himpi.venc_send_jpege_frame is called. The output data is directly stored in the buffer set by the user. Compared with the other mode, this mode does not require you to copy the output data from the buffer returned by DVPP to the specified memory. However, the JPEG header of the output data may contain the COM comment field (whose length ranges from 4 bytes to 19 bytes), and thus the data length is larger.

Version

Input Buffer Allocation/Freeing API

Atlas inference product

During JPEGE image encoding, call acl.himpi.dvpp_malloc/acl.himpi.dvpp_free to allocate or free the input buffer on the device. The buffer lifetime is managed by the user.

The following two methods are supported:

Managed by DVPP: Call acl.himpi.venc_send_frame to send source images for encoding.
Correctly set the attribute value of hi_venc_chn_attr["venc_attr"]["buf_size"] When creating a channel by calling acl.himpi.venc_create_chn. For the parameter description, see hi_venc_attr.

In this mode, the JPEG header of the output data does not contain the COM comment field, and the data length is smaller. However, you need to copy the output data from the buffer returned by DVPP to the specified memory.
Managed by users: Call acl.himpi.venc_send_jpege_frame to send source images for encoding.
When creating a channel by calling acl.himpi.venc_create_chn, set hi_venc_chn_attr["venc_attr"]["buf_size"] to 0 (for details about the parameter description, see hi_venc_attr). Then call acl.himpi.venc_get_jpege_predicted_size to estimate the output memory size, and call acl.himpi.dvpp_malloc or acl.himpi.dvpp_free to allocate or free the output memory.

In this mode, the output buffer address is set when acl.himpi.venc_send_jpege_frame is called. The output data is directly stored in the buffer set by the user. Compared with the other mode, this mode does not require you to copy the output data from the buffer returned by DVPP to the specified memory. However, the JPEG header of the output data may contain the COM comment field (whose length ranges from 4 bytes to 19 bytes), and thus the data length is larger.

Atlas A2 training product / Atlas A2 inference product

Atlas 200I/500 A2 inference product

Atlas A3 training product / Atlas A3 inference product

During JPEGE image encoding, the following two types of buffer allocation/freeing APIs are supported:

Allocate or free the input buffer on the device by calling acl.rt.malloc/acl.rt.free. The buffer lifetime is managed by the user.
Allocate or free the input buffer on the device by calling acl.himpi.dvpp_malloc/acl.himpi.dvpp_free. The buffer lifetime is managed by the user.

Note: The memory allocated by calling the acl.himpi.dvpp_malloc API is dedicated for media data processing. However, the address space of the dedicated memory is limited. If memory planning is concerned or memory resources are limited, you are advised to call the acl.rt.malloc API to allocate memory.

The following two methods are supported:

Managed by DVPP: Call acl.himpi.venc_send_frame to send source images for encoding.
Correctly set the attribute value of hi_venc_chn_attr["venc_attr"]["buf_size"] When creating a channel by calling acl.himpi.venc_create_chn. For the parameter description, see hi_venc_attr.

In this mode, the JPEG header of the output data does not contain the COM comment field, and the data length is smaller. However, you need to copy the output data from the buffer returned by DVPP to the specified memory.
Managed by users: Call acl.himpi.venc_send_jpege_frame to send source images for encoding.
When calling the acl.himpi.venc_create_chn API to create a channel, set hi_venc_chn_attr["venc_attr"]["buf_size"] to 0 (for details about the parameter description, see hi_venc_attr). Then call the acl.himpi.venc_get_jpege_predicted_size API to estimate the output buffer size, and call the acl.himpi.dvpp_malloc\acl.himpi.dvpp_free API to allocate or free the output buffer. (You can also call the acl.rt.malloc\acl.rt.free API to allocate or free the output buffer on the device.)

In this mode, the output buffer address is set when acl.himpi.venc_send_jpege_frame is called. The output data is directly stored in the buffer set by the user. Compared with the other mode, this mode does not require you to copy the output data from the buffer returned by DVPP to the specified memory. However, the JPEG header of the output data may contain the COM comment field (whose length ranges from 4 bytes to 19 bytes), and thus the data length is larger.

Restrictions on Image Formats, Width and Height Alignment, and Buffers

For details about the definition of the input image format, see hi_pixel_format. For details about the concepts such as width stride and height stride, see Terminology.

**Table 1** Restrictions on the image format, width and height alignment, and buffer size
Input Format	Input Width and Height	Input Width Stride, Height Stride, and Buffer Size	Output Format
YUV420SP NV12 8bit	Width: Must be a multiple of 2. Height: Must be a multiple of 2.	Width stride: Round up the width to the nearest multiple of 16 or any multiple of 16, for example, 128 (which offers better performance). The height stride does not need to be configured. Buffer size (in bytes) = Width stride x Height x 3/2	Output format: JPEG, for example, *.jpg Only Huffman coding is supported. Arithmetic coding and progressive coding are not supported.
YUV420SP NV21 8bit
YUV422Packed YUYV 8bit	Width: Must be a multiple of 2. Height: No alignment requirement.	Width stride: Double the width and then round up the result to the nearest multiple of 16. The height stride does not need to be configured. Buffer size (in bytes) = Width stride x Height
YUV422Packed UYVY 8bit
YUV422Packed YVYU 8bit
YUV422Packed VYUY 8bit

Parent topic: VENC/JPEGE APIs