I like your positive thinking, Bill.

R
On Apr 1, 2012 8:53 AM, "Nathan Rusch" <[email protected]> wrote:

>   Still recovering from last night here. What’s your excuse? ;)
>
> -Nathan
>
>
>  *From:* Bill Gilman <[email protected]>
> *Sent:* Saturday, March 31, 2012 10:43 PM
> *To:* Nuke user discussion <[email protected]>
> *Subject:* Re: [Nuke-users] global proxy format setting?
>
> But of course…  please forgive my California-centric solipsism, and here's
> a little gem from my time down at Weta:
>
> "When you're working 7 days, every night is a Friday night!"
>
>  On Mar 31, 2012, at 10:25 PM, Ron Ganbar wrote:
>
>  It's Sunday morning here, which is working day here... :-)
>
> R
> On Apr 1, 2012 7:42 AM, "Bill Gilman" <[email protected]> wrote:
>
>> What are all you people doing responding to this thread on a Saturday
>> night?!?
>>
>> This is what's wrong with VFX, right here.  ;-)
>>
>> I agree that every approach has it's own set of problems.  We're doing 4K
>> stereo with LOTS of iterations and creative changes in versions.  It's more
>> important for us to get rough iterations than finished work, at least at
>> this stage.  I'm sure there will be version issues, but we're trying to
>> take the step to the next level as a facility so we're trying to
>> standardize basic parts of the pipeline.  Using proxies is kind of a live
>> QA of our system in other ways.  Truth be told, as soon as something looks
>> too soft to the wrong person, I'm sure we'll have to bump everything up to
>> 4K.  'Til then, however…
>>
>>
>>
>>  On Mar 31, 2012, at 9:23 PM, Nathan Rusch wrote:
>>
>>   Hugo: It’s definitely easy to initially set such a system up from a
>> development standpoint, but when you start getting into situations where
>> versions of comp input sequences (mattes, paint work, lighting/fx renders,
>> etc.) are iterating rapidly and concurrently, having to ensure that all the
>> input sequences are synchronized to their local counterparts is more
>> trouble than it’s worth.
>>
>> Even if, for example, you forcefully disable proxy mode when a comp
>> renders on the farm, you can still run into edge cases where the artist has
>> been working in proxy mode, but what they have been looking at is outdated,
>> so you end up potentially wasting a fairly significant chunk of farm time.
>> Now in theory this SHOULDN’T happen if you keep a properly-versioned output
>> pipeline, but that doesn’t mean it won’t, and things like this ALWAYS go
>> wrong in crunch. Plus, since artists are working on multiple shots and will
>> likely switch between 3 or 4 before lunch, they almost certainly won’t
>> remember what versions of everything were synced when for which shots.
>>
>> Having been there and back (at a small, maneuverable studio, no less),
>> this is based on my experiences, so obviously, different people will have
>> different approaches and opinions. Because central storage is the pretty
>> much the most important part of your studio, it just seems like a better
>> investment to put money into improving its performance and network layout
>> than to spend developer time and money maintaining a compromise/workaround
>> solution. Simple things like avoiding layered EXRs and other
>> performance-averse formats like TIFF, managing compression and bit-depth
>> settings intelligently to make good use of existing storage, and tuning
>> your network filesystem appropriately can all go a long way.
>>
>>
>> Ron:
>> os.path.realpath(os.path.join(nuke.root()['project_directory'].value(),
>> read['file'].value()))
>>
>>
>> -Nathan
>>
>>
>>  *From:* Ron Ganbar <[email protected]>
>> *Sent:* Saturday, March 31, 2012 8:52 PM
>> *To:* Nuke user discussion <[email protected]>
>> *Subject:* Re: [Nuke-users] global proxy format setting?
>>
>>  Basically you are all talking about a custom Read node that is tied to
>> your project / sequence / shot setup and that has all these functions built
>> in.
>> In that case a lot of this becomes vert practical. The Read node can make
>> local copies and use them if they exists, and read from the Network if they
>> don't. I used a setup like this several times (even back in Shake days) and
>> it works very well.
>> On a small studio scale, I just never saw anyone put the effort in to do
>> this properly. This is when it becomes a pain in the backside.
>>
>> Incidentally, in case you are using the Project Directory path in the
>> Project Settings panel, then use a ./rest/of/my/path/seq.%04d.dpx how do
>> you resolve this easily via python?
>>
>>
>> Ron Ganbar
>> email: [email protected]
>> tel: +44 (0)7968 007 309 [UK]
>>      +972 (0)54 255 9765 [Israel]
>> url: http://ronganbar.wordpress.com/
>>
>>
>>
>> On 1 April 2012 06:43, Hugo Leveille <[email protected]> wrote:
>>
>>>  How do you find it impractical? Having a script that copies the
>>> network file to a local raid and set it in proxy path is quite practical
>>> and easy to manage, not matter how many artist, no? Unless I misunderstand
>>> you.
>>>
>>> Id like to have your view on this.
>>>
>>> Thanks
>>>
>>> Sent from my iPhone
>>>
>>> On 2012-03-31, at 11:30 PM, "Nathan Rusch" <[email protected]>
>>> wrote:
>>>
>>>    Yeah, localizing files is definitely nice (preferable to proxies
>>> like you said), but it becomes rapidly impractical once you exceed a
>>> certain number of artists.
>>>
>>> -Nathan
>>>
>>>
>>>  *From:* Hugo Leveille <[email protected]>
>>> *Sent:* Saturday, March 31, 2012 8:24 PM
>>> *To:* Nuke user discussion <[email protected]>
>>> *Subject:* Re: [Nuke-users] global proxy format setting?
>>>
>>>  What you'll want more often than other is a 1:1 local version of a
>>> network file. Not a mixed resolution
>>>
>>> Sent from my iPhone
>>>
>>> On 2012-03-31, at 11:20 PM, "Nathan Rusch" <[email protected]>
>>> wrote:
>>>
>>>    Once you cross that nebulous critical point where you go from using
>>> proxies occasionally to regularly, chances are you’re thinking about
>>> writing scripts/tools to handle it automagically.
>>>
>>> In your example, it would be possible to parse out the resolution string
>>> from the image path, but the easier choice would be to simply hand Nuke the
>>> path to the proxy sequence and let it set the format for you (using
>>> read['proxy'].fromUserText('/my/proxy/path')). This would also let you to
>>> work with a more generalized directory structure. An extremely simplified
>>> example would be something like:
>>>
>>>
>>> /pfcluster/Alphaville/PRODUCTION/01_joyluck/RENDERS/3D_ELEMENTS/joyluck_lightTex_v019_ln/LaserGlow/
>>> *full*/joyluck_lightTex_v019_ln_LaserGlow.####.exr
>>>
>>>
>>> /pfcluster/Alphaville/PRODUCTION/01_joyluck/RENDERS/3D_ELEMENTS/joyluck_lightTex_v019_ln/LaserGlow/
>>> *proxy*/joyluck_lightTex_v019_ln_LaserGlow.####.exr
>>>
>>> This way you won’t need to worry about doing any parsing, and can just
>>> set the proxy knob if that directory exists, or leave it blank otherwise.
>>>
>>> To be honest, however, proxies are enough of a pain that the benefits
>>> rarely outweigh the potential for mixups and mistakes, even at 4k and above.
>>>
>>> -Nathan
>>>
>>>
>>>  *From:* Bill Gilman <[email protected]>
>>> *Sent:* Saturday, March 31, 2012 6:39 PM
>>> *To:* Nuke user discussion <[email protected]>
>>> *Subject:* Re: [Nuke-users] global proxy format setting?
>>>
>>>  Hi Hugo
>>>
>>> First off, thanks for the code.  It doesn't seem to be working for me
>>> but maybe it will after you get a chance to play with it.
>>>
>>> Out of curiosity, how hard would it be to parse, say, a resolution
>>> directory and change it to the proxy res?  Would this be best done with TCL
>>> or Python?  eg.:
>>>
>>>
>>> /pfcluster/Alphaville/PRODUCTION/01_joyluck/RENDERS/3D_ELEMENTS/joyluck_lightTex_v019_ln/LaserGlow/
>>> *4096x3112*/joyluck_lightTex_v019_ln_LaserGlow.####.exr
>>>                             ^ ^ ^
>>>
>>> would become
>>>
>>>
>>> /pfcluster/Alphaville/PRODUCTION/01_joyluck/RENDERS/3D_ELEMENTS/joyluck_lightTex_v019_ln/LaserGlow/
>>> *1024x778*/joyluck_lightTex_v019_ln_LaserGlow.####.exr
>>>                             ^ ^ ^
>>>
>>> But my real question is:  everybody has to do this with shots they're
>>> using proxies for?  That seems really nuts to me.  I can't imagine everyone
>>> has automatic scripts to do this sort of thing.
>>>
>>> Here's to it
>>>
>>> Bill
>>>
>>>  _______________________________________________
>>> Nuke-users mailing list
>>> [email protected], http://forums.thefoundry.co.uk/
>>> http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users
>>>
>>> ------------------------------
>>> _______________________________________________
>>> Nuke-users mailing list
>>> [email protected], http://forums.thefoundry.co.uk/
>>> http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users
>>>
>>>  _______________________________________________
>>> Nuke-users mailing list
>>> [email protected], http://forums.thefoundry.co.uk/
>>> http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users
>>>
>>>
>>> _______________________________________________
>>> Nuke-users mailing list
>>> [email protected], http://forums.thefoundry.co.uk/
>>> http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users
>>>
>>
>>
>> ------------------------------
>> _______________________________________________
>> Nuke-users mailing list
>> [email protected], http://forums.thefoundry.co.uk/
>> http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users
>> _______________________________________________
>> Nuke-users mailing list
>> [email protected], http://forums.thefoundry.co.uk/
>> http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users
>>
>>
>>
>> _______________________________________________
>> Nuke-users mailing list
>> [email protected], http://forums.thefoundry.co.uk/
>> http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users
>>
> _______________________________________________
> Nuke-users mailing list
> [email protected], http://forums.thefoundry.co.uk/
> http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users
>
>
>
> ------------------------------
> _______________________________________________
> Nuke-users mailing list
> [email protected], http://forums.thefoundry.co.uk/
> http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users
>
> _______________________________________________
> Nuke-users mailing list
> [email protected], http://forums.thefoundry.co.uk/
> http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users
>
_______________________________________________
Nuke-users mailing list
[email protected], http://forums.thefoundry.co.uk/
http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users

Reply via email to