VTM_CTU_partition

Save and Dispaly VTM CTU partition info.

Johnny_1280x720_60_0.yuv, Johnny_1280x720_60_VTM.h266 and CTU/CTU_x.txt is only used for test.

This Project Usage:

First, install numpy, PIL, matplotlib for your Python3.

And then:
git clone https://github.com/codec2021/VTM_CTU_partition.git

python3 Draw_CTU_Partition.py

You will get the result picture:

Following the steps below, you can also obtain your own block partitioning results.

Step1. Change VTM22.0 Encoder Code

VTM Code is here: https://vcgit.hhi.fraunhofer.de/jvet/VVCSoftware_VTM.git

At the end of function void EncCu::compressCtu in EncCu.cpp file, Please add the following code:

std::string ctuNum = std::to_string(ctuRsAddr);
std::ofstream ctuPartitionFile;

std::cout << ctuNum << std::endl;
ctuPartitionFile.open("./CTU_" + ctuNum + ".txt");

for (auto &currCU : cs.traverseCUs(CS::getArea(cs, area, ChannelType::LUMA), ChannelType::LUMA))
{
	const CompArea&  lumaArea = currCU.block(COMPONENT_Y);
	int cuX = lumaArea.x;
	int cuY = lumaArea.y;
	int cuH = lumaArea.height;
	int cuW = lumaArea.width;
	std::string cuInfo = "";

	cuInfo = std::to_string(cuX) + " " + std::to_string(cuY) + " " + std::to_string(cuH) + " " + std::to_string(cuW) +"\n";
	ctuPartitionFile << cuInfo;
}
ctuPartitionFile.close();

Then build VTM22.0 Project to get the EncoderApp binary

Step2. run VTM Encoder

Prepare your test yuv and cfg file, then run

for example:

./EncoderApp -c ./encoder_randomaccess_vtm.cfg -i Johnny_1280x720_60.yuv -wdt 1280 -hgt 720 -fr 30 -f 1 -q 30 -b Johnny_1280x720_60_VTM.h266

you will get some file named as CTU_0.txt, CTU_1.txt ... file in your local dir.

Step3. move all CTU_x.txt file to local ./CTU Dir

mkdir CTU
mv *.txt ./CTU

Step4. run the Draw_CTU_Partition.py by python3

set the encoded yuv name, width, height, ctu_size in the Draw_CTU_Partition.py:

and then run:

python3 Draw_CTU_Partition.py