Thanks A.J. for going through all that trouble and posting the results of your tests. I'm not accustomed to read MTF in that scale but I can try to figure it out.

Looking at your pics (not MTF curves) it does look like the 20X HD lens is better in the center and that the 3x SD lens is not quite as good in the center but definitely better in the corners than the HD lens... I'm honestly not impressed with the corner performance of any of them. I should shut up since I still have to look closer.

Ash, I guess I come from the tradition of spending 50%-100% of the cost of the camera on a lens, so it's just expected. I'd rather have a mid range camera with a good lens than a high end camera with a cheap one. Of course the Sony F330 XDCAM and Grass Valley Infinity have me tempted since they're both using larger chips and are going to be around $25K with a basic lens.

If I bought a Varicam I'd have to spend at least $25K on a lens. If I buy an H1 I'd spend $5K or more on one. If you compare the cost ratio between camera & lens in both those packages you'll notice they're pretty even.

The whole point of owning a wide angle lens is to use it at the widest setting. I'd much prefer to own a fixed prime 3mm lens that is sharp, than a zoom that is soft at 3.4mm and sharp at 10.4mm.

When I use the Optex or Red Eye wide angle lens on the 16X Manual servo lens, I keep it at the fixed widest setting, and then use my body/feet/tripod to change any framing.

Most of us are quite happy with the original Canon 14X & 16X black manual lenses - and they seem to provide good results with the XL1/2/H1 bodies - so I'm sure if Canon made a 3X Manual Servo lens (at a decent price) with improved sharpness at the 3.4mm end, it would be snapped up by a lot of XL camera owners.

A sharp wide angle prime would be sweet.

A.J. not for criticism, but it's interesting to (non scientifically) compare those numbers to sharpness in Super 16mm film.

According to your chart, while the H1 (with manual lens) is initially maintaining a higher response as the cycles increase (to about 175 pph) it drops off faster to below 50% at 375 pph. Super 16 Vision 2 7218 negative film (with good lens) maintains 100% response to about 20 cycles/mm (144pph) and drops off slower to 50% response at around 70 cycles/mm (504 pph). For reference, I'm comparing the blue layer of the emulsion. Same stock in ("professional") Super 8 (with Zeiss glass) on the other hand starts to drop from 100% response at 80pph to 50% response at 280pph. Of course, on film, granularity plays in the game as well.

Those numbers are most interesting. The 8 and 16 mm numbers are consistent i.e. go to a frame of about half the size and get about half the resolution pph. They also give a sense of what a tremendous acheivement the XLH1 is, especially at the price. It delivers resolution comparable to Super 16 with a sensor size comparable to an 8 mm frame.

More interesting, I think, is the observation of faster rolloff and what the explanation for that might be. It occured to me while thinking about this that there have been no complaints about the XLH1 WRT jaggies/moire/aliasing and while they can be seen if you look for them they are certainly not as obtrusive as they were with the earlier XL cameras. Thus I suspect that the camera may contain an anti aliasing filter whereas it is clear that the XL2, for example, did not. The fact that the MTF's go cleanly to 0 at the folding frequency (540 cy) is indicative that the aliasing is minimal. Given that an antialising filter is present the decision to incorporate sharpening (idicated by the bumps in the MTF ant about 150 cy) seems the smart thing to do.

All this is speculative, of course.

Yeah, you're right it could be the anti aliasing filter... it could also be affected by electronic detail that's enhancing edges at the higher cycles and covering actual detail up. In other words, the edge enhancement around small details might be creating an electronic edge that covers the transition from one cycle to the other.