Originally Posted by JonStatt
A few people have claimed that when they went from a poorly converged projector to a fantastic one that everything looked sharper. But if you ever sit down on your sofa, and intentionally move one of the colours, you will notice very little, if any, perceived difference in sharpness if the convergence remains within 1-2 pixels.
It depends how close you sit, at 1.0x times seating width distance, 1 pixel off on RED is too much (excluding the fact we can correct for 1 pixel off), especially for HTPC use. Giving generic assessments is hard because of people's varying seating distances. You are not only trying to generically assess what others' may or may not notice at different seating distances, but also with different levels of vision. Combine those two factors and we have a complicated issue. Let's assume most people have near 20/20 or a step below...
2 Pixels off on RED at my 1.0 X seating distance is a complete joke (we are talking uncorrected), as everything becomes completely blurry.
Though I will agree 1/2 pixel is hard to see, it depends what you are doing and what you are watching. It is VERY hard to see if you do a back and forth, but it is much easier if you A/B two projectors of the same make/model and show them side-by-side. The truth is convergence of 1/2 to 1 pixel on RED depends how far off blue conv and green CA also combine and it creates a muckiness. I'm working on a convergence grading formula for a FUTURE version of the calculator, if any of you guys have comments on the weighting, I'm thinking like this:
The main adjustment in a future formula will account for seating distance, this assumes a seating distance of about 1.1x to 1.2x screen width.Red
= 15 points lost per 1 pixel off, 1.5 per 1/10th pixel off, or 0.3 points lost for each 5 point screen test.Blue
= 5 points per 1 pixel off, 0.5 points lost per 1/10th pixel, 0.1 points lost per 1/10th pixel at 5 point screen testGreen
= 10 points lost per per 1 pixel, 1/10th point per 1/10th pixel
Hence, lets say the RED is off 1 full pixel on average across the screen at a 5-point screen test but others were near spot on, my grade to you would be 85 (B).
Let's say BLUE was off 1 full pixel while RED was off 1/3rd pixel, my grade to you would be 90 (A- ... Very Good).
Let's say Red was off 1 pixel and blue 1/2 pixel, my grade would be 82 (B- ... Average to Acceptable, "Generally a keeper")
Let's say Red was off 1.5 pixels and blue and green were neglible, my grade would be 78.5 (C+ ... Below Average to Acceptable, "Might consider replacing")
Let's say Red was off .5 pixels and Green had CA error .5 pixels and Blue was off .25 pixels, grade is 89 (B+, Good)
Let's say Blue was off 1/2 pixel and Red was off 1/10th pixel, my grade is 97 (A+ Exceptional)
Truthfully the formula would probably be better if it were more strict beyond 1 pixel of any given color on an exponential scale, but I'm just messing around with some generic numbers right now, I'll fix it up after some testing I need to do. This is my generic baseline which will be adjusted later after more testing.Edited by coderguy - 5/20/13 at 6:20am