how to make a the gif in the main page? #6
Comments
|
Hello,
As you may have seen, the one showing on the toolbox is just a
visualization. I guess that's why you closed the issue?
I am working on the 3D representation, we actually have one working
example, but it depends on the calibration. The calibration itself has just
been fniished and sent as a paper. Soon it should be released. The
unfortunate problem is that the calibration is in matlab (because of a
collaboration with other institute) while the rest will still be in python.
Do you need a 3D metric representation (to match between point clouds) or
just one point cloud and it does not matter if it's metrically consistent?
Do you have a calibration for your camera?
…On Tue, 12 May 2020 at 18:14, lvgeng ***@***.***> wrote:
Closed #6
<#6>.
—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub
<#6 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AIFOPI7CNG4DDKSOL73VQJ3RRFYV3ANCNFSM4M6CIYKQ>
.
|
I am more interested in a metrically consistent representation. Maybe I am not smart enough, sometime I feel I lost in this toolbox.... |
|
It's not about being smart, it is about the knowledge and the framework.
The toolbox is a work in progress from one person, it is not a finished
product from a company. For as much as I put effort, it won't be perfect. I
am sorry for that, but I am just a student as you trying to help, I have my
limits and I make my errors. That's why I need other people like you to try
to understand what is being doing and try to contribute also.
I mean, it would also be nice to know what are you doing, how you want to
do it and a bit of background, who you are and stuff like this may make it
easier for me to help.
In order to have a fruitful discussion we need to understand our framework,
or what you know/think/understand and what I know/think/understand,
otherwise we won't understand each other, I am afraid. We can also have a
call if you need.
It's very hard to explain a complicated problem in a few lines in a message
or mail or comment, so please let's try to be clear about what we are
talking about.
Unfortunately, the term *calibration *has been used quite vaguely for the
plenoptic camera. One way could be for me to differentiate between *MLA
Calibration *, (MLA = microlens array) so finding the MLA centers and being
able to divide the raw image into a lot of microlenses images and *Metric
Calibration *that is relating the disparity information to a 3D point in
the world in meter coordinates. At the moment, the Toolbox only relies on
the *MLA Calibration *so some values and relationship have to be inferred
and they may not work for all cmaeras. Doing a full *Metric
Calibration *requires
more effort, taking picture of a checkerboard, and computations. I am
working on that, but it takes time. It won't come in one day. We do not
have an arXiv, but the idea of calibrating a camera is not something new,
you can read for example here (https://ieeexplore.ieee.org/document/7151596,
http://publications.lightfield-analysis.net/JHGP13_gcpr_inm.pdf,
https://www.pf.bgu.tum.de/pub/2016/zeller_co_stilla_isprs16b_pap.pdf,
http://openaccess.thecvf.com/content_iccv_2017/html/Nousias_Corner-Based_Geometric_Calibration_ICCV_2017_paper.html)
and there are even many more.
If you are interested in a metrically consisted representation, you need to
calibrate the camera first and then once you know the parameters you can
project the disparity to a point cloud. Here there is also anotehr toolbox
for calibration
https://github.com/RViMLab/ICCV2017-plenoptic-camera-calibration
Does this help you?
Luca
…On Wed, 13 May 2020 at 18:58, lvgeng ***@***.***> wrote:
Hello, As you may have seen, the one showing on the toolbox is just a
visualization. I guess that's why you closed the issue? I am working on the
3D representation, we actually have one working example, but it depends on
the calibration. The calibration itself has just been fniished and sent as
a paper. Soon it should be released. The unfortunate problem is that the
calibration is in matlab (because of a collaboration with other institute)
while the rest will still be in python. Do you need a 3D metric
representation (to match between point clouds) or just one point cloud and
it does not matter if it's metrically consistent? Do you have a calibration
for your camera?
… <#m_1446354994196443310_>
On Tue, 12 May 2020 at 18:14, lvgeng *@*.***> wrote: Closed #6
<#6> <#6
<#6>>. —
You are receiving this because you are subscribed to this thread. Reply to
this email directly, view it on GitHub <#6 (comment)
<#6 (comment)>>,
or unsubscribe
https://github.com/notifications/unsubscribe-auth/AIFOPI7CNG4DDKSOL73VQJ3RRFYV3ANCNFSM4M6CIYKQ
.
I am more interested in a metrically consistent representation.
By "calibration" do you mean the xml calibration file for the raytrix
camera or it's something you designed for a dual camera system?
If you have an arxiv that I can read, it would be really helpful.
Maybe I am not smart enough, sometime I feel I lost in this toolbox....
—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub
<#6 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AIFOPIY24UCII72FZ7LFH53RRLGTLANCNFSM4M6CIYKQ>
.
|
|
I am reading. |
|
Yes, we can use that. But I did not do it yet. The output format from the Raytrix calibration is made for their API and their workflow, which is in some way different from what I am doing here. |
https://github.com/PlenopticToolbox/PlenopticToolbox2.0/blob/master/ANIMATIONS/animation.gif
Hi there I am curious that how did you make this vissualization.
Is it rendered as different views from image rendering?
Or, if it is a real 3D representation (like a point cloud?), is there a detailed example for it?
The text was updated successfully, but these errors were encountered: