[Sidefx-houdini-list] Camera model

Jordi Bares Dominguez jordibares at gmail.com
Wed Nov 28 12:59:43 EST 2012


Agreed, it is not that this particular approach is a silver bullet, just trying to revisit every single part of the process.

:-)

jb

On 28 Nov 2012, at 17:32, Pablo Giménez <pablogipi at gmail.com> wrote:

> 2012/11/28 Jordi Bares Dominguez <jordibares at gmail.com>
> 
>> 
>> On 28 Nov 2012, at 11:35, Pablo Giménez <pablogipi at gmail.com> wrote:
>> 
>>> 2012/11/28 Neil Scholes <neil at uvfilms.co.uk>
>>> 
>>>> I would certainly love having a lens shader where you can import lens
>>>> distortion models from apps like 3DEqualiser etc.
>>>> 
>>>> rather than having to render larger resolutions than needed and
>> distorting
>>>> in nuke.
>>>> 
>>> Well, this is  debatable.
>>> If you "bake" the lens distortion in the render then you are forced to
>>> rerender in case for  soem reason this lens dostortion option changes in
>>> the future, or changes for some shots.
>> 
>> True, but the whole process is simplified as you don't need to deal with
>> various sets of transformations on various points in the pipeline… this may
>> be debatable, true, but I rather keep it simple and re-render when that
>> happens which is not that common in my experience.
>> 
>>> And this happens sometimes.
>>> Whereas rendering some extra pixels is always a good thing,
>> 
>> I do not agree with that, I think that is a waste of rendering for every
>> single pass of every single shot and you count the extra hours you probably
>> would be horrified about it as I am.
>> 
> Well, it depends on the scenario, for film, rendering 10% more pixels
> usually is not a big deal.
> But it is quite common to have shots that are really hard to render an due
> to changes in how it is going to be composited they needs to be rendered
> again and this cost more time/resources that rendering 10% more pixels in a
> bunch of shots.
> As always it depends on the scenario, being able to make the same thing in
> both render and comp is always good and welcome, so I don't have nothing
> against having a better camera model, just the opposite.
> Doing everything at the render stage is not always the best option, same
> thing goes about trying to "fake" as most as possible in comp.
> So a compromise between them depending on the situation is usually the best
> option.
> 
>> 
>>> just an
>>> overscan 1.1 will  allow you to add lens distortion in comp and also be
>>> sure that further comp manipulations like camera shakes will have enough
>>> pixels to work with.
>> 
>> I would rather have 3D camera shakes if possible… sometimes this may not
>> be feasible for many reasons but...
>> 
>> jb
>> 
>>> I think it is better to render a bunch of extra pixels as a rule rather
>>> than be forced to rerender a expensive render.
>>> 
>>>> 
>>>> 
>>>> 
>>>> 
>>>> Neil Scholes
>>>> 
>>>> +44(0) 7977 456 197
>>>> www.uvfilms.co.uk
>>>> 
>>>> 
>>>> 
>>>> 
>>>> 
>>>> 
>>>> On 28 Nov 2012, at 10:17, Jordi Bares Dominguez wrote:
>>>> 
>>>>> The moment you enter into physically base rendering you will end up
>>>> wanting to use physically correct shaders and that ultimately will take
>> you
>>>> to realise that is actually not that useful unless you get it all right
>>>> (physically correct camera and lights too) and also HDR balanced
>> properly
>>>> (no grading)
>>>>> 
>>>>> The cascade effect is huge and adapting to the approach is not easy
>> plus
>>>> takes longer to render each frame but in my experience is more than
>>>> acceptable for the things you gain.
>>>>> 
>>>>> Ultimately, like apple did when they designed the iPad, one single rule
>>>> forced a ton of changes everywhere (no pen, just your finger), with PBR
>> is
>>>> the same and certain render engines are already there or thereabouts,
>>>> specially Maxwell, Vray and Mental Ray alongside all the other flavours
>>>> inspired by these.
>>>>> 
>>>>> I really hope after this exchange of ideas SideFX see the value and put
>>>> together a physically correct ecosystem of tools for modern rendering.
>>>>> 
>>>>> my 2 cents
>>>>> jb
>>>>> 
>>>>> 
>>>>> On 28 Nov 2012, at 10:04, Neil Scholes <neil at uvfilms.co.uk> wrote:
>>>>> 
>>>>>> Interesting thoughts
>>>>>> 
>>>>>> I was thinking that rendering linear light values in high dynamic
>> range
>>>> is the real benefit  we already have, because after that all real camera
>>>> limitations such as abberation grain and iso can be faked in a more
>>>> painterly way in the comp which is surely quicker and easier ?
>>>>>> 
>>>>>> Also the issues of real world lenses and ISO etc well each camera,
>> lens
>>>> is going to vary in accuracy even if the camera manufacturers offer a
>>>> template - rather like f stop vs T stop.
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> Neil Scholes
>>>>>> 
>>>>>> +44(0) 7977 456 197
>>>>>> www.uvfilms.co.uk
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> On 28 Nov 2012, at 08:50, Jordi Bares wrote:
>>>>>> 
>>>>>>> I would love that, including the camera transform matrix please.
>>>>>>> 
>>>>>>> Jb
>>>>>>> 
>>>>>>> Sent from my iPhone
>>>>>>> 
>>>>>>> On 28 Nov 2012, at 00:32, Colin Doncaster <colin.doncaster at gmail.com
>>> 
>>>> wrote:
>>>>>>> 
>>>>>>>> 
>>>>>>>> Sure!  I would assume it would be helpful if all of the camera
>>>> parameters ended up in metadata stored in the output image - this fits
>> well
>>>> into the EXR camera info as it's meant to represent what exposure the
>> data
>>>> represents with a good reference point from where to adjust it.
>>>>>>>> 
>>>>>>>> This would make a nice COP and/or MPlay addition where you can
>>>> quickly adjust the values and round robin them back to the scene camera.
>>>>>>>> 
>>>>>>>> Cheers
>>>>>>>> 
>>>>>>>> On 2012-11-27, at 6:17 PM, Jordi Bares Dominguez <
>>>> jordibares at gmail.com> wrote:
>>>>>>>> 
>>>>>>>>> I don't imagine anyone aspiring to get noise simulation baked in
>>>> their render and with the ever expanding electronic cameras
>> configurations
>>>> and codecs this simply may be impossible but it may be useful to get a
>>>> metric for density of noise in the picture that we can feed into the
>> comp.
>>>>>>>>> 
>>>>>>>>> hope it makes sense
>>>>>>>>> jb
>>>>>>>> 
>>>>>>>> _______________________________________________
>>>>>>>> Sidefx-houdini-list mailing list
>>>>>>>> Sidefx-houdini-list at sidefx.com
>>>>>>>> https://lists.sidefx.com:443/mailman/listinfo/sidefx-houdini-list
>>>>>>> _______________________________________________
>>>>>>> Sidefx-houdini-list mailing list
>>>>>>> Sidefx-houdini-list at sidefx.com
>>>>>>> https://lists.sidefx.com:443/mailman/listinfo/sidefx-houdini-list
>>>>>>> 
>>>>>> 
>>>>>> _______________________________________________
>>>>>> Sidefx-houdini-list mailing list
>>>>>> Sidefx-houdini-list at sidefx.com
>>>>>> https://lists.sidefx.com:443/mailman/listinfo/sidefx-houdini-list
>>>>> 
>>>>> _______________________________________________
>>>>> Sidefx-houdini-list mailing list
>>>>> Sidefx-houdini-list at sidefx.com
>>>>> https://lists.sidefx.com:443/mailman/listinfo/sidefx-houdini-list
>>>>> 
>>>> 
>>>> _______________________________________________
>>>> Sidefx-houdini-list mailing list
>>>> Sidefx-houdini-list at sidefx.com
>>>> https://lists.sidefx.com:443/mailman/listinfo/sidefx-houdini-list
>>>> 
>>> 
>>> 
>>> 
>>> --
>>> Un saludo
>>> Best Regards
>>> Pablo Giménez
>>> _______________________________________________
>>> Sidefx-houdini-list mailing list
>>> Sidefx-houdini-list at sidefx.com
>>> https://lists.sidefx.com:443/mailman/listinfo/sidefx-houdini-list
>> 
>> _______________________________________________
>> Sidefx-houdini-list mailing list
>> Sidefx-houdini-list at sidefx.com
>> https://lists.sidefx.com:443/mailman/listinfo/sidefx-houdini-list
>> 
> 
> 
> 
> -- 
> Un saludo
> Best Regards
> Pablo Giménez
> _______________________________________________
> Sidefx-houdini-list mailing list
> Sidefx-houdini-list at sidefx.com
> https://lists.sidefx.com:443/mailman/listinfo/sidefx-houdini-list




More information about the Sidefx-houdini-list mailing list