Maxwell Render | Official public forum

what is 2 points perspective...?

Using photography terms... the request is for a Tilt/Shift lens for the camera. very important for architectural images to correct the converging vertical lines of buildings.

it basically corrects parallax when the camera and its target are not at the same height.

example:

I would also like to see this feature implemented.

n

you're right... but because I use a normal cam to take the background, they are 3 points pers. (where I'll comp the render ), I have to undistord the picture anyway, so I bought a photoshop plugin that is pretty good and does exactly that... now if U have some nice shots taken with an Hasselblad or similar... then maybe you're right, having that in Maxwell directly is cool...

just one cent...

the "point" in 2 point perspective of course is the vanishing 'points' The converging lines will usually converge on the horizon line, that is, if the object is parallel to the groundplane. All reality is 3point (min). A cube in any space has three sets of parallel converging lines. Only fake, or orthographic rendering can have 2 point, which means one of the dimensions of the cube is parallel and never converges to a 'point'.

A lot of beginner illustrators ignore the Y axis, and make it appear parallel, which is why most people start by learning 2 point perspective in drawing, even though it's not technically accurate, but usually lines on the Y axis converge at the nadir or zenith and are very difficult to conceptualize or even draw.

most complex objects like chairs, or cars have hundreds of converging lines... so 234 point perspective if you will..

So, it's not technically accurate to say 2point perspective if what you really want to get across is orthographic rendering on the Y axis... tho it IS easier to say.



BTW: Herve, what the heck IS that thing... a sheepdog? What's next? an upside-down sheepdog.. hehe

Alan, in the meantime, you can of course reduce the effect of convergence by moving the camera further away and zooming in, which reduces perspective, and will make the render look more orthographic. Lightwave doesn't even have a native Orthographic renderer, (that I know of) so I doubt a reality based render will adopt it... though it seems a lot of architects are using this for previs... so, I guess it would be helpful.

rivoli wrote:it basically corrects parallax when the camera and its target are not at the same height.

yes that would be very handy!! but there is always the trick to keep the camera leveled (no up/down inclination), and set the format the way to include all the scene elements you want to render (in this case the building). there will be a huge amount of foreground in the camera view window. then, crop render only the upper part which is of interest. works perfectly.

markus

here, as an example:



the cropped part is the red frame.

cheers

markus

Hey Ian, this is (was) a picture of James Clerk Maxwell dog... I am pretty sure they retouch the eyes... he he..

wow electric image still alive!

well this is very similar what cinema 4d does, but it does it automatical, no cropping needed. it has a yshift and a x shift parameter in the camera. also the xshift is very practica for getting the right perspective. i think this is the best done method to achieve architectural perspectives. nextlimit should take a look at the cinema 4d solution.

cheers
stefan

lllab wrote:wow electric image still alive!

well this is very similar what cinema 4d does, but it does it automatical, no cropping needed. it has a yshift and a x shift parameter in the camera. also the xshift is very practica for getting the right perspective. i think this is the best done method to achieve architectural perspectives. nextlimit should take a look at the cinema 4d solution.

cheers
stefan

electric image has it too, it's named camera shift and tilt. it was just a screenshot to illustrate the trick

markus

Since the Form-Z plugin is due in a few weeks I've been considering purchase, but the lack of two-point perspective is a big negative. form-Z has this capability and I use it almost always - they call it "keep vertical lines straight."

I understand that keeping the camera level, moving it back, and then cropping can be a workaround. But it seems that it would create much longer render times due to the additional image area and the necessity of rendering greater pixel resolution to compensate for so much of the image later discarded by cropping.

If Maxwell allows for a defined area of a view to be rendered (in effect cropping prior to the render) it may eliminate the longer render times. Can Maxwell do this?

Richard

There is a bitmap region render option.. so it should work

I don't have 2 points perspective in 3dsmax.