dzungdoan6 / g2d Goto Github PK

Hey, is it possible to extract segmentation masks for GTA 5 from G2D. I know its not written in the use cases but is it something that can be done by modifying the codebase?
Thanks

Hi again. I'm sorry for disturbing you so much, i wanted to ask you if there is some way to get images while playing the game i.e. Realtime. I tried your gdiscreencapture function but it just freezes the screen and then takes screen shots. i tried to take screenshots for 10 seconds but I am only getting multiple images of same frame as the game freezes on initialization of gdiscreencapture function. Please help.

I am running a cracked version of gta5 and after creating the vertices using the tool i cant find the vertex.txt file in the gta5 directory. Both the weather and trajectory tool are working but the file cannot be found i dont know if it is even created. Could you please help me with this. Thanks

Impressing work! But it seems that you did not mention how the camera intrinsics were achieved from the game.

Are you simply feed an initial guess to the SFM framework and let the algorithm optimizes it, or calculate it from information like FOV achieved from the game?

Hello @dadung,

In my work, basically I want to do simple night time restoration task: enhancing the brightness(making dark images brighter). For this, I need dataset:
input: night time images as provided by G2D
output: Same night time images with longer exposure, so that they look brighter.

Basically is there a way to adjust exposure settings while capturing the dataset.

Also, G2D provides foggy images, is there a way to adjust different fog densities and get the dataset with different fog levels(low fog, medium fog, severe fog as like adjusting traffic density from 0 to 1).

If any help, regarding above would be highly appreciated.

Thank you
Gopi

Hey!
First of all, congrats for your awesome job!
I was wondering, is it possible to change the camera agent and choose a car instead of a pedestrian?
Is there a way to automate such a process?

Thank you!

Hi, sorry to disturb you again i wanted to know whats the pose annotation for the camera, the x, y, z and are the rotations in degrees or radians and whats the positive direction for rotation and movements? Thanks in advance

Hi,

Thanks for the amazing work. Is it possible to extract depth maps (without using SFM)? If yes, can you please guide me on how to modify your code to do that?

Hi I would like to calibrate the gta5 camera that you have used for image extraction. I want to use the images for some other purpose for which i need the intrinsic parameters of the camera. Can you guide me, how i can get those features? Many Thanks

Hi, I am doing a fundamental matrix check on the images and poses provided by the code through epipolar lines, however when i check the epipolar lines they do not seem to pass through the selected object. The same check has been run on the Minos simulator and they provide accurate checks as the poses are also accurate. However on the poses provided here the check fails. As you can see the object i have selected in the first image is the railing on the left of the image and in the second image the epipolar line should pass through the same point in the railing but it doesn't. Please any help will be appreciated.

Thank you for releasing this amazing work!

One question. Can it be extended to collect frames captured in Mod like Free Cam Mode instead of the view of a running pedestrian?

Hi this is more a question than an issue does the game need to be bought from steam or can this work for a cracked game? thanks