Color. Obj file

[cwo75]

Thank you for the quick andere Titel my Demo_obj question!!!!!!!! I have 512x512x3 RGB values for my surface but when in put it to write_obj_color I get an error. ( if you need more info I send you the source Thank you Christian

[JBKacerovsky]

Hi christian, from the information you gave it sounds like you are applying uv coordinates and a texture file. Is that correct? While I think that would be a more elegant solution in line with .obj best practices that is not how the current version of the function is built. I tried to make the script similar in terms of input and usage to the built in patch function. So the color input requires either a vector of a continuous variable, Nx1 matrix and a colormap (one of the built in Matlab colormaps) or a list of RGB values, Nx3 matrix, where N should be equal to the number of faces or vertices of the mesh object. Each unique colour is saved as a different material. I know this is not super efficient, since it can lead to a potentially very large number of materials on the same object (and it doesn't really allow for smooth colours), but this was most straightforward to implement as a direct adaptation from the patch function, which I was using in Matlab before.

Does this answer your question? If you want to use this function, I think it should be relatively straightforward to pull out the list (Nx3 matrix) of RGB values from the colormap file you seem to have using uv coordinates. If you still run into issues (or if I misunderstood the question) please provide some example code and data.

Benjamin

[cwo75]

Reply Requested When Convenient

Hy!

It time for me to introduce myselfe to you I'm working for University of Linz, Johannes Kepler Universität Austria. I'm working on a project ( i try to cannot get all the resources from university witch is in a half corona lock down)

I try to visualisze US Data visualize to HMD and CAVE. my code:

A = imread ('Bild_100.png'); %(A 512x512) colormap jet;

fvc=surf2patch(X,Y,A,A)

%result % fvc.faces fvc.vertices fvc,facesvertexcdata

obj_write_color(fvc, 'demo', c);

lenght matches n vertices -> using vertex color median as face mtl definition Error using .* Integers can only be combined with integers of the same class, or scalar doubles.

Error in linspace (line 44) y = d1 + (0:n1).*(d2 - d1)./n1;

Error in obj_write_color (line 130) cbin_edges=linspace(cmin, cmax, RGB_bins+1);

Sorry i dont know why getting this error

mfg

christian

"J. Benjamin Kacerovsky" notifications@github.com 12.02.2021 22:22 >>> Hi christian, from the information you gave it sounds like you are applying uv coordinates and a texture file. Is that correct? While I think that would be a more elegant solution in line with .obj best practices that is not how the current version of the function is built. I tried to make the script similar in terms of input and usage to the built in patch function. So the color input requires either a vector of a continuous variable, Nx1 matrix and a colormap (one of the built in Matlab colormaps) or a list of RGB values, Nx3 matrix, where N should be equal to the number of faces or vertices of the mesh object. Each unique colour is saved as a different material. I know this is not super efficient, since it can lead to a potentially very large number of materials on the same object (and it doesn't really allow for smooth colours), but this was most straightforward to implement as a direct adaptation from the patch function, which I was using in Matlab before. Does this answer your question? If you want to use this function, I think it should be relatively straightforward to pull out the list (Nx3 matrix) of RGB values from the colormap file you seem to have using uv coordinates. If you still run into issues (or if I misunderstood the question) please provide some example code and data. Benjamin — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub ( https://github.com/JBKacerovsky/objWriter/issues/2#issuecomment-778462132) , or unsubscribe ( https://github.com/notifications/unsubscribe-auth/ASEMQWREBSOVCURVZNI6ZX3S6WLZFANCNFSM4XRD25WA) .

[JBKacerovsky]

Hi Christian,

Oh cool, ich komm' auch aus Österreich :) I studied in Vienna.

I took a look at your example. The error you are getting seems to be because different numeric types are passed to linspace. fvc.facevertexcdata from surf2patch on an 8-bit png is of numeric type uint8, but the other variables passed to linspace in line 130 are of type double, so it throws an error.

You can simply convert your color vector to double and it should fix the issue. I also added a quick fix into the code (simple enforcing types double for the relevant variables) so if you pull the repo it should be fixed and you should be able to run the obj_write_color.m function without having to convert anything.

Please note that the function is written specifically for triangle meshes, so you will have to specify 'triangles' for surf2patch (see code example below). If you pass quad-meshes in, some bizarre shapes emerge :p

below you have a code example (using one of the .png images that come with the Matlab distribution I believe) that shows the fix with the 'old' and new versions of the function:

`A=imread('peppers.png'); B=A(:, :, 1); [X, Y] = meshgrid(1:512, 1:384); %% X, Y as pixel coordinates

Z=rescale(mean(A, 3), 0, 90); %% Z scale as averagee rgb values

fvc=surf2patch(X,Y,Z,B, 'triangles');

% test with matlabb figure figure patch(fvc, 'edgecolor', 'none', 'facevertexcdata', double(fvc.facevertexcdata), 'facecolor', 'flat'); colormap jet axis equal

%% export colour obj file % this fixes the issue of the previous iteration of the function % the numeric typee of 'colors' gets passed to cmin and cmax. The linspace % function requires all input variables to be of the same type, in this % case double. obj_write_color((fvc), 'demo', double(fvc.facevertexcdata), 'colorMap', 'jet', 'illum', 10);

%% with updated version (feb-15) of obj_write_color.m % the most recent update of obj_write_color.m converts cmin and cmax to % double, so it should also work without issue if the numeric type of % colors is 'uint8', as is the case in the above example. obj_write_color((fvc), 'demo_double', (fvc.facevertexcdata), 'colorMap', 'jet', 'illum', 10);

%% testing accessing the rgb values for each pixel directly from source file and mapping to mesh/.obj file yy = fvc.vertices(:, 1); xx = fvc.vertices(:, 2);

r = A(:, :, 1); r = r(sub2ind(size(r), xx, yy)); g = A(:, :, 2); g = g(sub2ind(size(g), xx, yy)); b = A(:, :, 3); b = b(sub2ind(size(b), xx, yy));

rgb = [r, g, b];

rgb = double(rgb)/255;

obj_write_color((fvc), 'demo_rgb', rgb, 'illum', 5); `

[cwo75]

Reply Requested When Convenient

Hi !

So you studiet in Vienna , as Austria is not very big :-) this is only 150 km from me. Kommst du aus Österreich oder hat du nur in Österreich studiert. Some realy nice Projekts about Science Computing (VSC) are going on there. The Data from the US Simulation were made with VSC Vienna.

THANK YOU for you help ,i try it right away in a few moments. Now i dont write to you on the Handy becouse as you saw the writing mistakes my handy always translates the word into german to some really strange words and i didnt notice that.

I give you some results as soon as i finish working OK

Thank you.

mfg

Christian

-- Johannes Kepler Universitaet -- Christian Wohlschlager -- Altenbergerstr.69 -- 4040 Linz Austria -- Tel: +43 732 2468 3897

• Mail:christian.wohlschlager@jku.at

"J. Benjamin Kacerovsky" notifications@github.com 15.02.2021 22:59 >>> Hi Christian, Oh cool, ich komm' auch aus Österreich :) I studied in Vienna. I took a look at your example. The error you are getting seems to be because different numeric types are passed to linspace. fvc.facevertexcdata from surf2patch on an 8-bit png is of numeric type uint8, but the other variables passed to linspace in line 130 are of type double, so it throws an error. You can simply convert your color vector to double and it should fix the issue. I also added a quick fix into the code (simple enforcing types double for the relevant variables) so if you pull the repo it should be fixed and you should be able to run the obj_write_color.m function without having to convert anything. Please note that the function is written specifically for triangle meshes, so you will have to specify 'triangles' for surf2patch (see code example below). If you pass quad-meshes in, some bizarre shapes emerge :p below you have a code example (using one of the .png images that come with the Matlab distribution I believe) that shows the fix with the 'old' and new versions of the function: A=imread('peppers.png'); B=A(:, :, 1); [X, Y] = meshgrid(1:512, 1:384); %% X, Y as pixel coordinates Z=rescale(mean(A, 3), 0, 90); %% Z scale as averagee rgb values fvc=surf2patch(X,Y,Z,B, 'triangles'); % test with matlabb figure figure patch(fvc, 'edgecolor', 'none', 'facevertexcdata', double(fvc.facevertexcdata), 'facecolor', 'flat'); colormap jet axis equal %% export colour obj file % this fixes the issue of the previous iteration of the function % the numeric typee of 'colors' gets passed to cmin and cmax. The linspace % function requires all input variables to be of the same type, in this % case double. obj_write_color((fvc), 'demo', double(fvc.facevertexcdata), 'colorMap', 'jet', 'illum', 10); %% with updated version (feb-15) of obj_write_color.m % the most recent update of obj_write_color.m converts cmin and cmax to % double, so it should also work without issue if the numeric type of % colors is 'uint8', as is the case in the above example. obj_write_color((fvc), 'demo_double', (fvc.facevertexcdata), 'colorMap', 'jet', 'illum', 10); %% testing accessing the rgb values for each pixel directly from source file and mapping to mesh/.obj file yy = fvc.vertices(:, 1); xx = fvc.vertices(:, 2); r = A(:, :, 1); r = r(sub2ind(size(r), xx, yy)); g = A(:, :, 2); g = g(sub2ind(size(g), xx, yy)); b = A(:, :, 3); b = b(sub2ind(size(b), xx, yy)); rgb = [r, g, b]; rgb = double(rgb)/255; obj_write_color((fvc), 'demo_rgb', rgb, 'illum', 5); — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub ( https://github.com/JBKacerovsky/objWriter/issues/2#issuecomment-779466243) , or unsubscribe ( https://github.com/notifications/unsubscribe-auth/ASEMQWU3BGOBYMQTA262X73S7GKLBANCNFSM4XRD25WA) .

[cwo75]

Reply Requested When Convenient

Hier

Ja ja ja working now first coding succesful ltell you later when full progress ist Doberan MFG Christian

"J. Benjamin Kacerovsky" notifications@github.com 15.02.21 22.59 Uhr >>>Hi Christian,

Oh cool, ich komm' auch aus Österreich :) I studied in Vienna.

I took a look at your example. The error you are getting seems to be because different numeric types are passed to linspace. fvc.facevertexcdata from surf2patch on an 8-bit png is of numeric type uint8, but the other variables passed to linspace in line 130 are of type double, so it throws an error.

You can simply convert your color vector to double and it should fix the issue. I also added a quick fix into the code (simple enforcing types double for the relevant variables) so if you pull the repo it should be fixed and you should be able to run the obj_write_color.m function without having to convert anything.

Please note that the function is written specifically for triangle meshes, so you will have to specify 'triangles' for surf2patch (see code example below). If you pass quad-meshes in, some bizarre shapes emerge :p

below you have a code example (using one of the .png images that come with the Matlab distribution I believe) that shows the fix with the 'old' and new versions of the function:

`A=imread('peppers.png'); B=A(:, :, 1); [X, Y] = meshgrid(1:512, 1:384); %% X, Y as pixel coordinates

Z=rescale(mean(A, 3), 0, 90); %% Z scale as averagee rgb values

fvc=surf2patch(X,Y,Z,B, 'triangles');

% test with matlabb figure figure patch(fvc, 'edgecolor', 'none', 'facevertexcdata', double(fvc.facevertexcdata), 'facecolor', 'flat'); colormap jet axis equal

%% export colour obj file % this fixes the issue of the previous iteration of the function % the numeric typee of 'colors' gets passed to cmin and cmax. The linspace % function requires all input variables to be of the same type, in this % case double. obj_write_color((fvc), 'demo', double(fvc.facevertexcdata), 'colorMap', 'jet', 'illum', 10);

%% with updated version (feb-15) of obj_write_color.m % the most recent update of obj_write_color.m converts cmin and cmax to % double, so it should also work without issue if the numeric type of % colors is 'uint8', as is the case in the above example. obj_write_color((fvc), 'demo_double', (fvc.facevertexcdata), 'colorMap', 'jet', 'illum', 10);

%% testing accessing the rgb values for each pixel directly from source file and mapping to mesh/.obj file yy = fvc.vertices(:, 1); xx = fvc.vertices(:, 2);

r = A(:, :, 1); r = r(sub2ind(size(r), xx, yy)); g = A(:, :, 2); g = g(sub2ind(size(g), xx, yy)); b = A(:, :, 3); b = b(sub2ind(size(b), xx, yy));

rgb = [r, g, b];

rgb = double(rgb)/255;

obj_write_color((fvc), 'demo_rgb', rgb, 'illum', 5); `

-- You are receiving this because you authored the thread. Reply to this email directly or view it on GitHub: https://github.com/JBKacerovsky/objWriter/issues/2#issuecomment-779466243