sunset1995 / py360convert Goto Github PK
View Code? Open in Web Editor NEWPython implementation of convertion between equirectangular, cubemap and perspective. (equirect2cube, cube2equirect, equirect2perspec)
License: MIT License
Python implementation of convertion between equirectangular, cubemap and perspective. (equirect2cube, cube2equirect, equirect2perspec)
License: MIT License
I am trying to create a photosphere using a combination of a few normal camera images.
I know the fov, u, v and also the height and width of the input image.
My initial image (equirectangular)
My input image (planer or perspective view [fov=75, u=0, v=0, img_w=640, img_h=480])
I was able to get the (x,y) position using your e2p
function and fit those images in the sample equirectangular image near the equator(mid of the image). But as I go near the edges, the images start to overlap since I am missing the pixel manipulation to form curves.
Can you please provide some reading material, since your code has no documentation or, if you got time, can you please help me out.
I came across a tiny mistake in c2e
conversion with the 'dict' cube_format
and I wanted to share in case it would be helpful (I could make a pull request instead if desired).
In cube_dict2h()
in utils.py
, I think the variables in the assertion should be face_k
instead of k
:
def cube_dict2h(cube_dict, face_k=['F', 'R', 'B', 'L', 'U', 'D']):
assert len(face_k) == 6
# instead of:
# assert len(k) == 6
return cube_list2h([cube_dict[k] for k in face_k])
I also came across something that might be a mistake. In equirect_facetype()
also in utils.py
, it uses np.bool
, which doesn't seem to exist. Instead I specified the datatype as a boolean:
def equirect_facetype(h, w):
'''
0F 1R 2B 3L 4U 5D
'''
tp = np.roll(np.arange(4).repeat(w // 4)[None, :].repeat(h, 0), 3 * w // 8, 1)
# specify the dtype as boolean:
mask = np.zeros((h, w // 4), dtype='?')
# this gives me an error (it doesn't look like np.bool exists):
# mask = np.zeros((h, w // 4), np.bool)
idx = np.linspace(-np.pi, np.pi, w // 4) / 4
idx = h // 2 - np.round(np.arctan(np.cos(idx)) * h / np.pi).astype(int)
for i, j in enumerate(idx):
mask[:j, i] = 1
mask = np.roll(np.concatenate([mask] * 4, 1), 3 * w // 8, 1)
tp[mask] = 4
tp[np.flip(mask, 0)] = 5
return tp.astype(np.int32)
Thanks for this great tool :)
Unfortunately I cannot execute it, any idea what I'm doing wrong?
I tried both python 2.7 and python3, both fail:
> python -m convert360 --convert e2c --i equirectangular-0.jpg -o e2c.png --w 512
D:\Program Files\Python27\python.exe: No module named convert360
> python -m py360convert --convert e2c --i equirectangular-0.jpg -o e2c.png --w 512
Traceback (most recent call last):
File "D:\Program Files\Python27\lib\runpy.py", line 151, in _run_module_as_mai
n
mod_name, loader, code, fname = _get_module_details(mod_name)
File "D:\Program Files\Python27\lib\runpy.py", line 109, in _get_module_detail
s
return _get_module_details(pkg_main_name)
File "D:\Program Files\Python27\lib\runpy.py", line 101, in _get_module_detail
s
loader = get_loader(mod_name)
File "D:\Program Files\Python27\lib\pkgutil.py", line 464, in get_loader
return find_loader(fullname)
File "D:\Program Files\Python27\lib\pkgutil.py", line 474, in find_loader
for importer in iter_importers(fullname):
File "D:\Program Files\Python27\lib\pkgutil.py", line 430, in iter_importers
__import__(pkg)
File "D:\Program Files\Python27\lib\site-packages\py360convert\__init__.py", l
ine 1, in <module>
from .e2c import e2c
File "D:\Program Files\Python27\lib\site-packages\py360convert\e2c.py", line 3
, in <module>
from . import utils
File "D:\Program Files\Python27\lib\site-packages\py360convert\utils.py", line
68
out = np.ones((*out_hw, 3), np.float32)
^
SyntaxError: invalid syntax
> python3 -m py360convert --convert e2c --i equirectangular-0.jpg -o e2c.png --w 512
C:\Users\me\scoop\apps\python\current\python.exe: No module named py360convert
> python3 -m py360convert --convert e2c --i equirectangular-0.jpg -o e2c.png --w 512
C:\Users\me\scoop\apps\python\current\python.exe: No module named py360convert._
_main__; 'py360convert' is a package and cannot be directly executed
> python3 -m convert360 --convert e2c --i equirectangular-0.jpg -o e2c.png --w 512
C:\Users\me\scoop\apps\python\current\python.exe: No module named convert360
Btw, with this tool, how can I generate 6 different image files for the cubemap faces?
Just wondering how difficult it would be.
I am using p360convert with python 3.7.8. I tried e2c and it worked as expected but I am having issue in c2e. I used the example, but same issue appeared. Am I doing it wrong or is this a bug?
cube_dice = np.array(Image.open('assert/example_e2c.png'))
Image.fromarray(py360convert.c2e(cube_dice,400,800), 'RGB').show()
Is there a way to get a mapping between coordinates of equirectangular and cubemap image
When I play the code on Python it gives me this error and I think it is a problem with PIL
/bin/python3 "/home/jubei/Documentos/Python/Images 360/equiretangular.py"
Traceback (most recent call last):
File "/usr/lib/python3/dist-packages/PIL/Image.py", line 2813, in fromarray
mode, rawmode = _fromarray_typemap[typekey]
KeyError: ((1, 1, 3), '<f8')
The above exception was the direct cause of the following exception:
Traceback (most recent call last):
File "/home/jubei/Documentos/Python/Images 360/equiretangular.py", line 9, in <module>
image = Image.fromarray(equiretangular)
File "/usr/lib/python3/dist-packages/PIL/Image.py", line 2815, in fromarray
raise TypeError("Cannot handle this data type: %s, %s" % typekey) from e
TypeError: Cannot handle this data type: (1, 1, 3), <f8
Thanks for your library @sunset1995 ! I was wondering if you could tag a new release for pypi; seems like some bugs have been fixed since the initial release. Thanks!
Hey,
Using your example code :
import numpy as np
from PIL import Image
import py360convert
cube_dice = np.array(Image.open('MyImage.jpg'))
cube_h = py360convert.cube_dice2h(cube_dice) # the inverse is cube_h2dice
cube_dict = py360convert.cube_h2dict(cube_h) # the inverse is cube_dict2h
cube_list = py360convert.cube_h2list(cube_h) # the inverse is cube_list2h
print('cube_dice.shape:', cube_dice.shape)
print('cube_h.shape:', cube_h.shape)
print('cube_dict.keys():', cube_dict.keys())
print('cube_dict["F"].shape:', cube_dict["F"].shape)
print('len(cube_list):', len(cube_list))
print('cube_list[0].shape:', cube_list[0].shape)
I've got this error message :
Traceback (most recent call last):
File "PhotoToCubeMap.py", line 8, in <module>
cube_h = py360convert.cube_dice2h(cube_dice) # the inverse is cube_h2dice
File "C:\Program Files\Python37\lib\site-packages\py360convert\utils.py", line 217, in cube_dice2h
assert cube_dice.shape[0] == w * 3 and cube_dice.shape[1] == w * 4
AssertionError
Thank you for this great contribution!
Yet, I am wondering where we could find are some reading materials that might help us understand the theory foundation behind the implementations.
Any recommendation would be appreciated!
In py360convert/utils.py line no: 193, you have an unknown variable reference "k" commenting that line works though.
def cube_dict2h(cube_dict, face_k=['F', 'R', 'B', 'L', 'U', 'D']):
# assert len(k) == 6
return cube_list2h([cube_dict[k] for k in face_k])
For photogrammetry it is helpful to have projected images with overlap on each face with its neighbouring faces.
With some rewriting to support Numba, this library could be even faster. It would require a bit of refactoring, but it might make the code even faster and unlocks the ability to use fastmath, autoparallelilzation and other features.
The jit_module function from numba gets you part of the way there, but there were several parts of the functions like c2e that needs to be rewritten into smaller functions or slightly modified to use the supported functions in numpy.
TODO:
In the README.md, c2e categorize the input cubemap's format, 'dice' (default), 'horizon', 'dict' and 'list'. But, there's no further explain about that format because i guess there's no formal sentence for cubemap format.
I searched about cubemap formap and there's some shape of cubemap.
So, if anyone have time to add the more precise explain about format. Please add it :)
Thank you for reading.
A declarative, efficient, and flexible JavaScript library for building user interfaces.
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google ❤️ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.