Re: [PyMOL] High Resolution Ray Tracing

[Stephen Kerry]

Hi Thomas,

Thank you for letting me know that this is a genuine limitation of the code as 
it stands. The strange thing is that despite the vast memory usage, the final 
PNG images are rarely over 10 MB, unless there are many lines from a CCP4 map 
for example. Making the scene from scratch rather than from a session file 
helped reduce memory usage during ray tracing, but still gave me the same 
segmentation fault.

Do you know a workaround to divide a scene into halves or quarters and process 
the individual segments separately? I could then combine the PNG images 
afterwards. However, I fear this might be harder to achieve than it sounds, as 
the lighting and ray shadows would have to match up across the segments to 
produce a seamless final image.

Cheers,

Stephen


From: Thomas Holder 
Sent: 12 May 2016 17:56:02
To: Stephen Kerry
Cc: pymol-users@lists.sourceforge.net
Subject: Re: [PyMOL] High Resolution Ray Tracing

Hi Stephen,

This is not an artificial Open Source PyMOL limitation.

PyMOL's ray tracing code was written at a time when probably nobody had a 
computer with 64GB of RAM, or at least didn't attempt to ray trace images that 
size. The code is pretty hard to read and understand, and currently there is 
nobody who's familiar with it and could easily fix such a limitation. I assume 
the code makes an assumption which is only true up to a certain memory address 
range.

Cheers,
  Thomas

On 12 May 2016, at 10:20, Stephen Kerry  wrote:

> So the images are back from the printers and I can easily tell the difference 
> between 180 and 300 dpi. You have to look closely to tell the difference 
> between 300 and 600 dpi. I am unable to distinguish between 600 and 1200 dpi.
>
> My image size is still limited to 7500 x 6000, which at 600 dpi gives a 12.5 
> x 10 inch image. Is this an artificial Open Source PyMOL software limitation 
> or a genuine bug/memory leak? The segmentation faults also seems to happen 
> even with 128 GB and 256 GB of RAM!
>
> On my system with 32 GB of RAM, I see PyMOL reserves more virtual RAM than 
> the amount of physical RAM, which might be why it crashes when trying to 
> write the PNG at the end of ray tracing? It does not appear to use this 
> virtual RAM or even compress any RAM, according to activity monitor on OSX 
> 10.10. Disabling memory compression and/or swap makes no difference.
>
> Is there anything else I can try to successfully output such a large image?
>
> 
> From: Thomas Holder 
> Sent: 10 May 2016 20:34:44
> To: harold steinberg
> Cc: Stephen Kerry; pymol-users@lists.sourceforge.net
> Subject: Re: [PyMOL] High Resolution Ray Tracing
>
> Hi Adam et al.,
>
> This is a super interesting and helpful discussion!
>
> Just wanted to throw in the following shortcut for creating the 6" test 
> images:
>
> png image1.png, 6in, ray=1, dpi=180
> png image2.png, 6in, ray=1, dpi=300
> png image3.png, 6in, ray=1, dpi=600
> png image4.png, 6in, ray=1, dpi=1200
>
> Cheers,
>  Thomas
>
> On 10 May 2016, at 14:27, harold steinberg  wrote:
>
>> The 9600 x 4800 is a very common poster printer spec. Most print shops use 
>> 180 dpi images on their poster printers (for best print speed) and customers 
>> cannot tell the difference between that and a higher resolution.
>>
>> As a test, render a small image (say 6” x 6”) in PyMOL at four different 
>> dpi, 180 dpi (ray 1080), 300 dpi (ray 1800), 600 dpi (ray 3600) and 1200 dpi 
>> (ray 7200). Then use Photoshop or GIMP to make them all 6” x 6”.
>>
>> Put them together side-by-side in a layout program and print them on one 
>> sheet to compare. I bet you will have an extremely difficult time telling 
>> them apart.
>>
>>> On May 10, 2016, at 12:40 PM, Stephen Kerry  
>>> wrote:
>>>
>>> Harold, I agree 2400 dpi might be wasted, but the professional printer 
>>> supports up to 9600 horizontal dpi x 4800 vertical dpi, so a source image 
>>> with 1200 dpi would be a nice step up from 600 dpi. The lpi is 1200 as the 
>>> four passes from each CMYK component apparently increases the vertical dpi 
>>> by four.
>>
>> H. Adam Steinberg
>> 7904 Bowman Rd
>> Lodi, WI 53555
>> 608/592-2366

--
Thomas Holder
PyMOL Principal Developer
Schrödinger, Inc.


--
Mobile security can be enabling, not merely restricting. Employees who
bring their own devices (BYOD) to work are irked by the imposition of MDM
restrictions. Mobile Device Manager Plus allows you to control only the
apps on BYO-devices by containerizing them, leaving personal data untouched!
https://ad.doubleclick.net/ddm/clk/304595813;131938128;j
___
PyMOL-users mailing list (PyMOL-users@lists.sourceforge.net)
Info Page: 

Re: [PyMOL] High Resolution Ray Tracing

[Thomas Holder]

Hi Stephen,

This is not an artificial Open Source PyMOL limitation.

PyMOL's ray tracing code was written at a time when probably nobody had a 
computer with 64GB of RAM, or at least didn't attempt to ray trace images that 
size. The code is pretty hard to read and understand, and currently there is 
nobody who's familiar with it and could easily fix such a limitation. I assume 
the code makes an assumption which is only true up to a certain memory address 
range.

Cheers,
  Thomas

On 12 May 2016, at 10:20, Stephen Kerry  wrote:

> So the images are back from the printers and I can easily tell the difference 
> between 180 and 300 dpi. You have to look closely to tell the difference 
> between 300 and 600 dpi. I am unable to distinguish between 600 and 1200 dpi.
> 
> My image size is still limited to 7500 x 6000, which at 600 dpi gives a 12.5 
> x 10 inch image. Is this an artificial Open Source PyMOL software limitation 
> or a genuine bug/memory leak? The segmentation faults also seems to happen 
> even with 128 GB and 256 GB of RAM!
> 
> On my system with 32 GB of RAM, I see PyMOL reserves more virtual RAM than 
> the amount of physical RAM, which might be why it crashes when trying to 
> write the PNG at the end of ray tracing? It does not appear to use this 
> virtual RAM or even compress any RAM, according to activity monitor on OSX 
> 10.10. Disabling memory compression and/or swap makes no difference.
> 
> Is there anything else I can try to successfully output such a large image?
> 
> 
> From: Thomas Holder 
> Sent: 10 May 2016 20:34:44
> To: harold steinberg
> Cc: Stephen Kerry; pymol-users@lists.sourceforge.net
> Subject: Re: [PyMOL] High Resolution Ray Tracing
> 
> Hi Adam et al.,
> 
> This is a super interesting and helpful discussion!
> 
> Just wanted to throw in the following shortcut for creating the 6" test 
> images:
> 
> png image1.png, 6in, ray=1, dpi=180
> png image2.png, 6in, ray=1, dpi=300
> png image3.png, 6in, ray=1, dpi=600
> png image4.png, 6in, ray=1, dpi=1200
> 
> Cheers,
>  Thomas
> 
> On 10 May 2016, at 14:27, harold steinberg  wrote:
> 
>> The 9600 x 4800 is a very common poster printer spec. Most print shops use 
>> 180 dpi images on their poster printers (for best print speed) and customers 
>> cannot tell the difference between that and a higher resolution.
>> 
>> As a test, render a small image (say 6” x 6”) in PyMOL at four different 
>> dpi, 180 dpi (ray 1080), 300 dpi (ray 1800), 600 dpi (ray 3600) and 1200 dpi 
>> (ray 7200). Then use Photoshop or GIMP to make them all 6” x 6”.
>> 
>> Put them together side-by-side in a layout program and print them on one 
>> sheet to compare. I bet you will have an extremely difficult time telling 
>> them apart.
>> 
>>> On May 10, 2016, at 12:40 PM, Stephen Kerry  
>>> wrote:
>>> 
>>> Harold, I agree 2400 dpi might be wasted, but the professional printer 
>>> supports up to 9600 horizontal dpi x 4800 vertical dpi, so a source image 
>>> with 1200 dpi would be a nice step up from 600 dpi. The lpi is 1200 as the 
>>> four passes from each CMYK component apparently increases the vertical dpi 
>>> by four.
>> 
>> H. Adam Steinberg
>> 7904 Bowman Rd
>> Lodi, WI 53555
>> 608/592-2366

-- 
Thomas Holder
PyMOL Principal Developer
Schrödinger, Inc.


--
Mobile security can be enabling, not merely restricting. Employees who
bring their own devices (BYOD) to work are irked by the imposition of MDM
restrictions. Mobile Device Manager Plus allows you to control only the
apps on BYO-devices by containerizing them, leaving personal data untouched!
https://ad.doubleclick.net/ddm/clk/304595813;131938128;j
___
PyMOL-users mailing list (PyMOL-users@lists.sourceforge.net)
Info Page: https://lists.sourceforge.net/lists/listinfo/pymol-users
Archives: http://www.mail-archive.com/pymol-users@lists.sourceforge.net

Re: [PyMOL] High Resolution Ray Tracing

[Stephen Kerry]

So the images are back from the printers and I can easily tell the difference 
between 180 and 300 dpi. You have to look closely to tell the difference 
between 300 and 600 dpi. I am unable to distinguish between 600 and 1200 dpi.

My image size is still limited to 7500 x 6000, which at 600 dpi gives a 12.5 x 
10 inch image. Is this an artificial Open Source PyMOL software limitation or a 
genuine bug/memory leak? The segmentation faults also seems to happen even with 
128 GB and 256 GB of RAM!

On my system with 32 GB of RAM, I see PyMOL reserves more virtual RAM than the 
amount of physical RAM, which might be why it crashes when trying to write the 
PNG at the end of ray tracing? It does not appear to use this virtual RAM or 
even compress any RAM, according to activity monitor on OSX 10.10. Disabling 
memory compression and/or swap makes no difference.

Is there anything else I can try to successfully output such a large image?


From: Thomas Holder 
Sent: 10 May 2016 20:34:44
To: harold steinberg
Cc: Stephen Kerry; pymol-users@lists.sourceforge.net
Subject: Re: [PyMOL] High Resolution Ray Tracing

Hi Adam et al.,

This is a super interesting and helpful discussion!

Just wanted to throw in the following shortcut for creating the 6" test images:

png image1.png, 6in, ray=1, dpi=180
png image2.png, 6in, ray=1, dpi=300
png image3.png, 6in, ray=1, dpi=600
png image4.png, 6in, ray=1, dpi=1200

Cheers,
  Thomas

On 10 May 2016, at 14:27, harold steinberg  wrote:

> The 9600 x 4800 is a very common poster printer spec. Most print shops use 
> 180 dpi images on their poster printers (for best print speed) and customers 
> cannot tell the difference between that and a higher resolution.
>
> As a test, render a small image (say 6” x 6”) in PyMOL at four different dpi, 
> 180 dpi (ray 1080), 300 dpi (ray 1800), 600 dpi (ray 3600) and 1200 dpi (ray 
> 7200). Then use Photoshop or GIMP to make them all 6” x 6”.
>
> Put them together side-by-side in a layout program and print them on one 
> sheet to compare. I bet you will have an extremely difficult time telling 
> them apart.
>
>> On May 10, 2016, at 12:40 PM, Stephen Kerry  
>> wrote:
>>
>> Harold, I agree 2400 dpi might be wasted, but the professional printer 
>> supports up to 9600 horizontal dpi x 4800 vertical dpi, so a source image 
>> with 1200 dpi would be a nice step up from 600 dpi. The lpi is 1200 as the 
>> four passes from each CMYK component apparently increases the vertical dpi 
>> by four.
>
> H. Adam Steinberg
> 7904 Bowman Rd
> Lodi, WI 53555
> 608/592-2366

--
Thomas Holder
PyMOL Principal Developer
Schrödinger, Inc.


--
Mobile security can be enabling, not merely restricting. Employees who
bring their own devices (BYOD) to work are irked by the imposition of MDM
restrictions. Mobile Device Manager Plus allows you to control only the
apps on BYO-devices by containerizing them, leaving personal data untouched!
https://ad.doubleclick.net/ddm/clk/304595813;131938128;j
___
PyMOL-users mailing list (PyMOL-users@lists.sourceforge.net)
Info Page: https://lists.sourceforge.net/lists/listinfo/pymol-users
Archives: http://www.mail-archive.com/pymol-users@lists.sourceforge.net

Re: [PyMOL] Contact map visualizer

[James Starlight]

Thanks Tsjerk!

now the map is accepted fine by the Pymol- however there is an problem
due to the mismatching of atoms in ref structure and the map (although
I use ref.pdf from which contact map has been produced). Probably the
source of the error is Martini's atom representation used in my system
- I've had the same issue in case of some VMD script which do the same
task.  Do any alternativeties are exist which can identify residues
(even on one of the protein) crucial for the formation of the complex
during md run with the possibility to map it to the structure for the
visualization?

J.

2016-05-12 12:52 GMT+02:00 Tsjerk Wassenaar :
> Hey :)
>
> That's been a choice of the author. You can contact him and ask.
>
> Cheers,
>
> Tsjerk
>
> On May 12, 2016 12:25, "James Starlight"  wrote:
>>
>> Ok, thanks!
>>
>> however it's normal that pymol ask me for the contact map in the png
>> and not for the original xpm produced by gromacs? It seems for the
>> that to operate directly with  Pymol outputs is more trivial :-)
>>
>> J.
>>
>> 2016-05-12 6:51 GMT+02:00 Tsjerk Wassenaar :
>> > Hi James,
>> >
>> > You can convert the .xpm file to .png/.jpg using tools like convert
>> > (imagemagick) and Gimp. Convert doesn't always get the Gromacs .xpm
>> > right,
>> > but it's an easy one to try.
>> >
>> > Cheers,
>> >
>> > Tsjerk
>> >
>> > On May 11, 2016 2:40 PM, "James Starlight" 
>> > wrote:
>> >>
>> >> Dear Pymol users!
>> >>
>> >> I am in charge with the analysis of protein-protein association during
>> >> long molecular dynamic simulation. In particularly I am interesting to
>> >> find residues on one of the protein which are crustal for the binding
>> >> interface established during Md.
>> >> For that purpose I am trying to use Contact map visualizer plugin to
>> >> map contact maps produced by gromacs onto the 3D structure via Pymol.
>> >> The problem that on the stage of loading of the contact map Pymol
>> >> proposed me to load only png or jpg files (not xpm produced by
>> >> g_mdmat). How it's possible to solve the issue? Any other suggestions
>> >> regarding my topic?
>> >>
>> >> Thanks!
>> >>
>> >> James
>> >>
>> >>
>> >>
>> >> --
>> >> Mobile security can be enabling, not merely restricting. Employees who
>> >> bring their own devices (BYOD) to work are irked by the imposition of
>> >> MDM
>> >> restrictions. Mobile Device Manager Plus allows you to control only the
>> >> apps on BYO-devices by containerizing them, leaving personal data
>> >> untouched!
>> >> https://ad.doubleclick.net/ddm/clk/304595813;131938128;j
>> >> ___
>> >> PyMOL-users mailing list (PyMOL-users@lists.sourceforge.net)
>> >> Info Page: https://lists.sourceforge.net/lists/listinfo/pymol-users
>> >> Archives: http://www.mail-archive.com/pymol-users@lists.sourceforge.net

--
Mobile security can be enabling, not merely restricting. Employees who
bring their own devices (BYOD) to work are irked by the imposition of MDM
restrictions. Mobile Device Manager Plus allows you to control only the
apps on BYO-devices by containerizing them, leaving personal data untouched!
https://ad.doubleclick.net/ddm/clk/304595813;131938128;j
___
PyMOL-users mailing list (PyMOL-users@lists.sourceforge.net)
Info Page: https://lists.sourceforge.net/lists/listinfo/pymol-users
Archives: http://www.mail-archive.com/pymol-users@lists.sourceforge.net