pdf from vcs looks different on OSX and Linux and or clients

[aashish24]

@doutriaux1 I found one issue is that the PDF generated seems to be bad.

The good news is that I found the source of the bad PDF problem and I am hoping that if we fix that, we could potentially fix the bad rendering of the pdf as well. The source of the problem is:

1. Text rendering: Somehow the labels are causing this issue as they inject something in PDF which is bad.

2. A pure VTK and GL2PS example didn't exhibit this issue so I am thinking that somehow we have a bad scene but need to confirm it.

3 I will push a branch which will turn of text rendering in PDF. I would be interested in knowing if the issue of different rendering still shows up @doutriaux1

[aashish24]

@doutriaux1 can you post a screenshot of how they looked different? It will help me a lot.

[aashish24]

Also, what SHA of VTK you were using?

[doutriaux1]

vtk from uvccdat channel 7.1.0.2.6 build uvcdat_master so I'm guessing the sha is v2.6 or 2.6

[aashish24]

thanks @doutriaux1

Good news: I found one problem and I am wondering if that's what causing the PDF to looked different.

Few questions:

1. Did the text (labels) look different or the data as well? or both?
2. Can you post screenshots if possible?

[aashish24]

test.pdf

@doutriaux1 @danlipsa could you guys download this PDF and visualize it in acroread and other pdf viewers and tell me if you still get the 1) error 2) look different?

[cdatrobot]

Works fine for me. No errors and looks the same.

On Fri, Nov 18, 2016 at 12:11 PM, Aashish Chaudhary < [email protected]> wrote:

test.pdf https://github.com/UV-CDAT/vcs/files/600503/test.pdf

@doutriaux1 https://github.com/doutriaux1 @danlipsa https://github.com/danlipsa could you guys download this PDF and visualize it in acroread and other pdf viewers and tell me if you still get the 1) error 2) look different?

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/UV-CDAT/vcs/issues/37#issuecomment-261586524, or mute the thread https://github.com/notifications/unsubscribe-auth/AS3dfCBdqojhz19-ZLuvcyFh8CgJa9t6ks5q_dxSgaJpZM4K06G8 .

[aashish24]

waiting for @doutriaux1

[aashish24]

Also, text as paths seems to be broken (which is the default right now).

[aashish24]

I have a partial fix but need to work with @dlonie (probably on Monday) to push a VTK branch.

[aashish24]

@doutriaux1 sorry to bother but could you read my previous messages?

[aashish24]

@doutriaux1 ping!

[aashish24]

@doutriaux1 could you please try this branch:

https://github.com/UV-CDAT/VTK/pull/16 (http://www.geuz.org/gl2ps)

If it works for you, I can clean it up and get it merged into UV-CDAT/vtk. I would have to get approval from GL2PS folks before I can merge it into proper GL2PS/VTK

[aashish24]

So far, I am seeing consistent results but at the same time, i am drawing very simple plots; I will try to create some more realistic plots and see if I can re-create the issue.

[doutriaux1]

@aashish24 trying now.

[doutriaux1]

@aashish24 this vtk branch along with @sankhesh PR?

[sankhesh]

@doutriaux1 I just posted the link to the VTK merge request on my PR.

[doutriaux1]

@sankhesh soI use yours instead of this @aashish24 VTK branch?

[aashish24]

@doutriaux1 you probably want to try my branch separately but you/I could merge my branch with Sankhesh branch...

[aashish24]

thanks for trying it out @doutriaux1 if you can send me your script, I can try it out and push a fix for the indexing error.

• aashish

[aashish24]

@doutriaux1 I fixed the silly bug on my end 😢 can you try the same branch again?

Thanks,

[doutriaux1]

building vtk branch as we speak

[aashish24]

Thanks. Looking at the PDF sankhesh generated, it looks correct on adobe pro but it does have issues on evince and acroread on Linux. The Chrome PDF renders is fast and correct. This PDF has my fixes so no more errors. I am looking into this issue some more and trying to compare it with CDAT 1.5 release.

[doutriaux1]

@scottwittenburg I think we might be able to close this I'll let you decide