[dyarza]

shift lenses are mounted in a housing which allows the lens to be moved laterally.

So the lens is shifted in relation to where the film is. Sounds just like offseting the film to me, I dont care if you move the lens or the film, the effect is the same.

As lllab states, this is something that is supported by most applications, and as you state, it can be already done in other ways. After all, it is just the same as rendering a bigger image and cropping a region of it. So my point still stands, it should be pretty darn easy to get M~R to read the values from the film offset box and render accordingly. We are not talking about distorting the lens module or anything like that.

Maybe to those that do not need such a function it does not seem like it is all that important so it is OK to have to use some workaround, but for those of us doing mainly Architectural rendering it is very important. I don't do much animation so maybe I should insist that NL don't waste their time integrating with the animation features of Maya... then again, that would be stupid.

[Ernesto]

I see there is a lot of confusion in this matter.

Mihnea Balta and Dyarza, are right, since a Shift Lens is exacltly the same as a Film Offset, and both are used to correct the perspective. All is relative, the important thing is that the axis of the lens is moved from the filmgate center. It does not matter if to do so, you move the lens or the filmback.
Maxwell has a virtual camera, so you could do both things since there are no phisical limitations.

Answering Mverta:

1) What you are asking was posted in the first messager in this same topic.
If you look carefully the image at the left (that says Maya Render or Mental Ray) was rendered with an Offset value set at 0.62, and the image at its right looks like using an offset setting of Zero.

2) Please note that you are wrong stating a diference between shift lens and film back offset. Both are the same, and both are intended to correct perspective.

3) I still wonder which is the method you are talking about, to get an image like the rendered withy Mental Ray. I would be more than gratefull if you could explain it in detail, so that I could reproduce it using Maxwell.
If what you are trying to say is modifiyng the FILM TRANSLATE values, I have just done a simple test, and I am sorry to tell it didn´t worked.

4) I suspect tat you are talking of a diferent matter...
Just to make this as clear as I can, and avoid any missunderstanding, I will write a short description:
I am trying to render the full 20 store building, in a way that vertical lines looks paralel and verticals in the rendered image, and at the same time the camera is at ground level, while the full building is shown in the image.
This camera is obviously pointing ahead, and up, because of the great height of the subject. I am posting a Maya file with a simplified version of the tower. This file was set using an offset value of 0.4 so that the building shows up uncropped.

http://www.4dempire.com/anima/offset.mb

[mverta]

Hi Mverta,shift lenses move the film behind the lense.

No, they don't.



The lens moves, not the film. Period. No, it's not the same thing, especially for Maxwell. And that's doubly true for next-gen Maxwell.


However...

You're going to hate me for this: I've been going crazy trying to understand why this was so difficult, especially since I was performing the offset using my image output size and getting the exact results in about 2 seconds. Shortly after requesting to see the camera attribute panel, I realized I'm running an additional plug-in which gives me this functionality, and it is not seen in commercial Maya. It has nothing to do with shift-lenses, obviously (it's for matte work), and I completely forgot that it was enabled. So it works for me; not for you. Sorry.

But the fact still stands that what we're talking about here is not shift-lens/perspective correction. It looks kinda like it though, but let's not pretend for .0002 seconds that Maya does two-point perspective. The workaround you're using, with film offset is close enough, I guess, but as Mihnea's stated, it's not implemented in Maxwell.

I understand this is an important feature for architects. You may not realize that Maxwell caters primarily to architects at this point, but it does. Perhaps this feature will accordingly be added one day.

But personally, as a vfx guy, I'm waiting for camera blur, render passes, animateable .mxi textures, parallel lights, and stacked materials before I can even begin to use Maxwell in animation/production work, so consider yourselves lucky

_Mike

[JDHill]

Guys,

This is not about 'moving the lens or the film', it's about throwing away some of what the lens 'sees'.

From the 'lens' point of view, nothing changes, but inside the camera, it's the same as shooting with 2x higher film...but instead, you just offset a standard-sized film and discarded the 'upper' or 'lower' information. The lens still 'sees' everything, you're just not capturing it due to the 'non-existent' part of the film. And that's why it's not just so simple for Maxwell...the very heart of Maxwell is about modeling the physical lens.

We are not talking about distorting the lens module or anything like that.

In fact, that's exactly what you're talking about...giving Maxwell the ability to only 'see' part of the scene. This is why doing a 2x-high region render is the physically-correct approach...and therefore already compatible with Maxwell. The 'feature' you're after is really being able to tell Maxwell, 'I want you to render 2x higher film, but throw away part of the output.', which is exactly what this type of camera does, and is also a technically accurate description of region render.

The upshot is, if/when this feature is implemented, it will not be anything you don't already have...the ability to discard a portion of the image before it's been rendered.

[dyarza]

The upshot is, if/when this feature is implemented, it will not be anything you don't already have...the ability to discard a portion of the image before it's been rendered.

Exactly, I keep saying this. So why is is difficult to implement? All I want is for M~R to discard a portion of the image before it is rendered, based on the values form the film offset box.

[Ernesto]

Mverta,

I was paying so much attention to what you were saying, that I thought you had a solution. I was expecting you to answer as an A team member, rather than as a user.

The answer that I was needing was:
Yes Ernesto you are righ, Mental ray works fine using the Offset option, but although in the Maxwell camera window you have the same boxes, they are not working.
This means that the Maxwell cameras are not that integrated into Maya as I was expecting.
The rest is only Useless Talk.

Mverta wrote:
"So it works for me; not for you. Sorry"

So, please give the solution to the NL team, so that we all can do it!!


E

[lllab]

mverta,
i just talked to the architectural photographer, he is a well know professional photographer for years here.

he still believes it is the same result, but how ever it is a thing urgently needed by us architects or at least render region for final output(which is basicly the same).

earnesto showed it very well.

you also brought up no solution, just wise words.

i really hope this will not be in "some" day, but very very soon. maxwell IS a very good tool for architects, but then it needs an architectural camera, without this it isnt a very good architectural tool.

by the way this is not a 2 point perspective. it is a special case of 3 point method where the 3rd point converges in infinity. this is only when you keep the camera absoluly horizonal. we had to study perpective for years in university. and i am also teaching this stuff now. believe me i know what i am talking about regarding perpective construction...

best
stefan

p.s.: the relation between film and lense is the thing - your third example is exactly what i meant and what cinema 4d does. it IS a shift lense. thanks for proving it in your diagramm. this is exactly what i meant and what we needed.
programming wise you get the same when you crop a virtual bigger image(or virtual "film plate"), we are just trying to implement that for vray.

[mverta]

Ernesto -

I didn't write the plug-in I use which gives me this functionality, but I have mentioned it to the higher-ups. Hope you get it.

_Mike

[joie]

Hey Mverta, is it possible to you to post that MEL here for us or is it secret?, I´d very glad to see what you say it does.

Thank´s

[lllab]

jdhill:
"This is not about 'moving the lens or the film', it's about throwing away some of what the lens 'sees'"

..yes exactly!
and i am sure this isnt too hard to implement, so please NL. would make maxwell even better with just this tiny addition. in studio preview it works already. please add this for mxcl-d renderings and alos netrender.

thanks very much!
Stefan

[mverta]

Hey Mverta, is it possible to you to post that MEL here for us or is it secret?, I´d very glad to see what you say it does.

Thank´s

It's not a mel, it's a plug-in from a very, uh, large company you've heard of So no... I want to work again, someday It basically works like the Canvas Size tool in Photoshop, where you can specify an anchor and it just extends the frame in the other directions. It's used for matte work, or situations where the camera absolutely can't move or change, etc.

_Mike

[Boris Ulzibat]

Hi Mverta,shift lenses move the film behind the lense.

No, they don't.



The lens moves, not the film. Period. No, it's not the same thing, especially for Maxwell. And that's doubly true for next-gen Maxwell.

I think you are not right. If two objects are aligned on an axis through their centers, and then they are shifted from being aligned with centers - it does not matter which of them IS shifted. It is like a glass half-full or half-empty.
Look at your picture, to me it looks like the film in the lower picture is shifted down from the center, to you it looks like the lens is shifted upwards, but in any case the result is the same - the centers of lens and film are not aligned - hence the changes in perspective.
It is just a matter of terminology, nothing more, and second, if Maxwell is said to mimic the real lens/camera/film system it must not be a problem to make this system "unaligned". Actually in any direction.
Please, correct me if i am wrong.

[lllab]

"It is just a matter of terminology"

exactly:-)
cheers
stefan

[joie]

What a pitty really Mverta...

[mverta]

In a virtual world, I guess you can say that the two yield the same result, so what's the difference - it's just semantics. But in reality, a lens shifting in its housing is not even remotely the same as film moving around in the gate. Film doesn't move around in the gate. Maxwell is based on, and will continue to be ever strictly based on, duplicating the conditions of reality, not virtual reality. So these sorts of distinctions are important. Plus, if you go around thinking that shift-lenses work by moving film around in the gate, you'll look like an ass.

I suppose some of the blame rests on Alias for naming it Film Offset if they meant Lens Offset, but that battle has long since been lost. In any case, it's not just a matter of terminology: moving film in the gate is different than moving a lens in the housing. On this planet, anyway. But obviously once virtual, you could name it anything you want. You could call it Image Realignment Modifier. Hell, you could call it Cheese Sandwich, what's the difference, right? The difference is that the Maxwell philosophy embraces real conditions, and real terminology. Plus, it's nice to have a clue.

_Mike