TypeError: set_camera_extrinsics(): incompatible function arguments.

Question

TypeError: set_camera_extrinsics(): incompatible function arguments.

ChristianMilianti opened this issue a year ago · comments

Hey Jad,
Really cool project!

I'm trying to run it on an iPhone 13 Pro Max, but I run into the following error when I try to send an image from my phone to instant-ngp when I run the following command $python scripts/nerfcapture2nerf.py --stream.

Do you know what the problem could be?

Cheers!
Christian

Jad Abou-Chakra · Answer 1 · Tue Mar 21 2023 16:39:11 GMT+0800 (China Standard Time)

Hey Christian, Problem should be solved after the latest merge in instantngp

…

On Tue, 21 Mar 2023 at 5:58 pm, Christian Milianti ***@***.***> wrote: Hey Jad, Really cool project! I'm trying to run it on an iPhone 13 Pro Max, but I run into the following error when I try to send an image from my phone to instant-ngp when I run the following command $python scripts/nerfcapture2nerf.py --stream. Do you know what the problem could be? [image: Screenshot from 2023-03-21 18-56-18] <https://user-images.githubusercontent.com/66559598/226546643-c4012131-238b-47e5-971d-1f6d708bbda0.png> Cheers! Christian — Reply to this email directly, view it on GitHub <#1>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AHCZMIXZ2EI63SQRD265KLTW5FNRTANCNFSM6AAAAAAWCCI4ZE> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

Christian Milianti · Answer 2 · Tue Mar 21 2023 17:27:29 GMT+0800 (China Standard Time)

Thanks for that Jad, that seems to have worked!
Another question I have is when I run

python scripts/nerfcapture2nerf.py --stream

the images are sent to instant-ngp and I can see them fine, but I'm not sure how to visualise the nerf that is being trained? On the other hand, If I first save the captured images using

python scripts/nerfcapture2nerf.py --save_path "dir1/dir2" --n_frames 20

and then drag and drop the saved folder into instant-ngp the nerf is created.

Do you know how to visualise the nerf directly when python scripts/nerfcapture2nerf.py --stream is run?

Jad Abou-Chakra · Answer 3 · Tue Mar 21 2023 17:54:09 GMT+0800 (China Standard Time)

Hey, if the gui is coming up, when first image comes in, are you starting the training by pressing “t” or by pressing the green button? Make sure the images are captured around the origin otherwise training wont work.

…

On Tue, 21 Mar 2023 at 7:27 pm, Christian Milianti ***@***.***> wrote: Thanks for that Jad, that seems to have worked! Another question I have is when I run python scripts/nerfcapture2nerf.py --stream the images are sent to instant-ngp and I can see them fine, but I'm not sure how to visualise the nerf that is being trained? On the other hand, If I first save the captured images using python scripts/nerfcapture2nerf.py --save_path "dir1/dir2" --n_frames 20 and then drag and drop the saved folder into instant-ngp the nerf is created. Do you know how to visualise the nerf directly when python scripts/nerfcapture2nerf.py --stream is run? — Reply to this email directly, view it on GitHub <#1 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AHCZMITMQL32WG345CXVYODW5FYAZANCNFSM6AAAAAAWCCI4ZE> . You are receiving this because you commented.Message ID: ***@***.***>

Christian Milianti · Answer 4 · Tue Mar 21 2023 18:02:45 GMT+0800 (China Standard Time)

By pressing the green button

Jad Abou-Chakra · Answer 5 · Tue Mar 21 2023 18:27:53 GMT+0800 (China Standard Time)

Could be the images are taken too far away from the unit box?

Christian Milianti · Answer 6 · Wed Mar 22 2023 07:36:57 GMT+0800 (China Standard Time)

Here's a video of what I mean (skip to about 20 sec in)

I can get the images to be sent from my phone to instant-ngp no problem using python scripts/nerfcapture2nerf.py --stream

But when I click start training, I'm not sure how to visualise the actual nerf that is being trained, I can only pan around the images that I've taken.

Whereas if I first save the images using python scripts/nerfcapture2nerf.py --save_path "dir1/dir2" --n_frames 20, and then simply drag and drop the saved folder, the nerf visualisation automatically appears.

nerf--stream-attempt1.mp4

Jad Abou-Chakra · Answer 7 · Wed Mar 22 2023 07:52:46 GMT+0800 (China Standard Time)

Oh I see, you’re in groundtruth mode. Press “G” to toggle back

…

On Wed, 22 Mar 2023 at 9:37 am, Christian Milianti ***@***.***> wrote: Here's a video of what I mean I can get the images to be sent from my phone to instant-ngp no problem using python scripts/nerfcapture2nerf.py --stream But when I click start training, I'm not sure how to visualise the actual nerf that is being trained, I can only pan around the images that I've taken. Whereas if I first save the images using python scripts/nerfcapture2nerf.py --save_path "dir1/dir2" --n_frames 20, and then simply drag and drop the saved folder, the nerf visualisation automatically appears. https://user-images.githubusercontent.com/66559598/226764910-ac618136-d3db-4ab0-81f4-af44d17c4522.mp4 — Reply to this email directly, view it on GitHub <#1 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AHCZMIWKVNDGG56EEDY3WB3W5I3SHANCNFSM6AAAAAAWCCI4ZE> . You are receiving this because you commented.Message ID: ***@***.***>

Christian Milianti · Answer 8 · Wed Mar 22 2023 08:04:23 GMT+0800 (China Standard Time)

Ahhhhh yes, it automatically defaulted to that it seems.

Thanks for your assistance, that worked! :)