DIY anamorphic lens

[dougr]

on the software correction for CA idea...

I know this is a bit old, but thought of this thread when I read it on the panamorph site:

"Unfortunately, LCA in the horizontal stretch case is so visible that some type of color correction is mandatory and all of today's popular HE lenses require extra lens elements to compensate for this (which is why HE lenses are so much more expensive than VC lenses). However, we have another trick we can play as well. LCA can be imagined as slight geometric scaling of each color component making up the image relative to some base color such as, for example, green. If the projector (or external scaler) can slightly vary the stretch of each color to compensate for the optical stretch on the screen, the remaining color aberration can be neutralized to within one half of a pixel. This patented technology has already been demonstrated. However, actual implementation into projectors and scalers is something only the market can decide. While the minor residual LCA of current vertical compression lenses is not enough to warrant this capability, it is possible that growth in the demand of a low-cost, high performance HE lens may provide enough motivation to develop it."

[Mark Techer]

I'm no expert on this however, I do see some limitations:
1. A single chip DLP can not have RGB displaced due to the design.
2. 3 chips sets like LCD and DILA can, however in just one direction.

Thoughts?

Mark

[noah katz]

"A single chip DLP can not have RGB displaced due to the design."

Shifting pixels in s/w will work for any design.

[Mark Techer]

A fair comment. It should noted that CA is a progressive abberation and not fixed to multiples of a pixel as well as it also bi-directional. This means it could be 1/10 of a pixel at 25% in from the edges and 2.6 pixels at the edges. Will S/W have that kind of flexibility?

I honestly belive that this can not be fixed electronically with the current projector technology.

Also, what happens if the user wants to move the lens for 16:9? Suddenly the screen is full of misconverged colours.

[noah katz]

No, but as was alluded to already, full per-pixel implementation could hold the error to no more than 1/2 pixel.

I'd consider myself lucky to have panel convergence that good on a 3-chip pj.

[dougr]

software could have that kind of flexibility... as long as the phenomena is well understood and consitent, i see it as no different than any other kind of image processing... in fact, it is easier than some. edge enhancement is very dynamic and requires user selectable parameters... certainly not a static pixel contrast displacement across the image (i.e. depends on where the high contrast edges are). same goes for algorithms to smooth motion. I think it can certainly be done, but the place for it to happen is in the image processing chip of the PJ, not externally. That way, when you select vertical stretching, the same logic that triggers the anamorphic lens moving into place in some systems will trigger a reconfiguration of the image processing much in the same way you have different calibration sets for different sources. I think panamorph is correct here (at least as far as they are concerned holding the patent); the market desire for such a solution (or lack thereof) will determine whether or not we will see it.

Reminds me of some technology I saw from a small company that sells stereoscopic machine vision systems (security systems, factory automation, etc.)... they had all the processing usually done on workstations on a dedicated chip... very complex algorithms to map stereoscopic images to a 3D model in high resolution/high update rate... they found a market for it, so putting on a chip became the better business model for them. The problem here, is that such a solution would likely be short lived, for if sufficient market existed to justify investment, we'd see 64:27 projectors shortly thereafter given the competitiveness of that marketplace.

[dougr]

come to think of it, such algorithms already are in widespread use in the photography world... there are many lens-specific photoshop (or other software) 'actions' to correct CA tailored to the actual lens and conditions of the photo... seems fairly straightforward to me technically.

[yuki]

hi``i am sduting zemax, and try to design a astro telescope, the software you are using now is also zemax?

Nice meet u!
you could add me or email me:
skype: yuki_hanxue
email: yuki.optic@gmail.com

[Mark Techer]

haha, oh the memories reading back over this thread

[PPW350]

Hi, I'm new to this forum, found it through browsing on AVS. Looking for a DIY A lens setup for my Optoma HD70. Can someone direct me to the kits that the members here are talking about?