First: My understanding of the term "vignetting," as used here at Lumenlab, is that it is a word used to describe this phenomenon: A "loss of light in the corners (of our projections)," or a "gain of light in the center" (of our projected images).

For the purposes of this discussion then, let's assume the following: We are talking about either: 1) dim corners (relative to the middle); or

2) A bright center (relative to the corners). AKA, a "hotspot."

Further, and for now, let's confine ourselves to just the latter--we'll assume the level of lighting in the corners is acceptable, but the level of lighting in the center is a bit over the top.

If the center of the image is "too bright," we can insert a filter to dim it down to match the level of lighting we see out in the corners. A customized "neutral density filter."

Such a filter would, in theory, be a "greyscale" filter...shaped rather like a "target." Or...if you prefer...shaped rather like the "center projected image" of any common fresnel lens.

In practice: one might simply print off (on clear celluloid) a greyscale image of said "target"....and then insert it behind (or in front of) the LCD in any of our projectors.

Darkest in the center...and progressively lighter as one moves further out toward the corners, assuming a rectangular filter....

Such a filter need NOT have a huge number of zones to be workable. Where a typical fresnel lens might sport 300 "rings," for example from center to edge, I suspect our "filter" might get away with just six "light density zones". Or twelve (we'll have to experiment to be confident).

The idea is, of course, to REDUCE the light to the center of the LCD, while keeping the level of light at the edges unchanged...

...therby evening out the overall level of lighting, and reducing (or even eliminating) any "hotspot"...and. thus, that effect known as "vingetting."

It seems to me such filters might easily be printed out on most home laser/inkjet printers, using celluloid (instead of paper) as a substrate.

Perhaps not a new idea, (certainly NOT a new idea with me, except, perhaps, that I've brought it over here...but I did search this forum for "neutral density filters" and came up blank).

In any case: I hope it helps.

k

Viable idea...I dunno about making it with rings....but a gradient might work better....It would require a large amount of testing to get the right tone/shade...And it would differ per each projector....But it sounds like a pretty good idea....

Well, it certainly would be a question of gradient. That's the whole idea (finding the proper gradient).

Such filters might reasonably be designed/manufactured using either: 1) math or 2) direct observation, using a lumenmeter placed in different areas on the screen (I gather a few members here have such equipment at their disposal).

Either method should at least give some basis for simple testing (and obviously, anything much beyond this is just fine-tuning).


k

Just a quick idea in case no one thought of it.

Given that these would be made on a case by case basis, then wouldn't it be much faster to say fire up something like Photoshop to make exactly the gradient you needed for the very PJ.

I.E. With the PJ all fired up do the gradient "live" so that when the image is viewed fullscreen with the PJ running you get even lighting.

Does that make sense or am I too tired at this point?

The ideas presented above have indeed been batted around in the past here but TTBOMK no one has ever bothered to implement them. The levels of vignetting we are able to achieve are apparently satisfactory to even the most discriminating eye so no one has bothered with dimming the projected image even further.

Actually yeah, that's a great idea. Run the projector with a computer, fire up photoshop in full screen mode, and design a gradient that, when shown full screen, evens things out.

That said, I'm against the whole idea. I say get the image as even as possible using the tools we have, and enjoy the bright center slightly more than the dark corners. Don't darken the center down to match.

The fact that 99% of the people with DIY projectors don't think they are bright enough would be enough to suggest they won't want 10-20% less light with a "filter".

Easiest solution:

Set up projector with no filter.

Take picture of "white" screen.

Invert colours on the photo. Adjust it so that the lightest parts (Darkest on projection) is white, with the rest being scaled down.

Print on transparency to size of LCD.

Place behind LCD. Admire your no vignetting projection.

Remove transparency to regain lost lumens.

But I think that assumes perfect (pixel for pixel) alignment between filter and LCD. Or at least perfect adjustability (and probably in all three planes).

Or am I missing something?

Not saying it couldn't be done, btw. but it would certainly be a challenge.


k

Actually, there is a good example of constructing just such a image-specific filter on the web. But it uses "the GIMP," rather than Photoshop.

http://www.gimpguru.org/Tutorials/NDFilter/

Assuming you were projecting a single image, I imagine you could easily print out just such a filter to slip into your projector.

But when discussing a filter for use with a moving image, it seems to me a "target" type filter would be most desirable.

Sounds like it's not a big issue, really, as many DIY builders might just prefer having a brighter image, overall, to a more evenly lit, but dimmer, projection.

Note too that it's not necessarily important to even the lighting all the way across the image--it might be enough simply to "even it up a bit more."

In any case, and for those who are battling this issue, it seems to me this is one possible solution.


k

Oh I don't think anyone is thinking it would be easy or that you would not lose some brightness. I just like these sorts of problems. Very abstract thinking required to solve them.

I may try Supras idea this weekend and see if it works.

Actually, what I'm suggesting does not require pixel for pixel alignment. A far more general approach is fine.

It would require pixel for pixel mapping if you were going to try to use Cyan, Magenta and Yellow filters on the Red, Green, and Blue sub-pixels, but we're not, a monochrome greyscale is more than adequate. I think, however, that there's better ways to acheive the same effect.

Re: Lack of pixel mapping.
Simply move the transparency away from the LCD slightly.