Kaku, Shoot a Rez chart if you can!!!
 

Kaku, in this HV200 section a thread was started on filming res charts along with links to pdf res charts.

Could you please film like 2 or 3 seconds of the charts in:

1080P/24- normal sharpenss at 0

1080P/24- sharpenss at -5

720P/24- normal sharpenss at 0

720P/24- sharpenss at -5

All in 0db gain too!

Use the highest res chart as well, either the 1920x1080 or even the higher res one.

Thanks Kaku.


res chart thread....
http://www.dvinfo.net/conf/showthread.php?t=57237

If you can, it might be best to post some uncompressed still frames for each of these cases (if you do compress, chose a lossless format such as TIFF LZW, ZIP, etc; and if the files are still too big, then try using JPEG at 100% quality as a last resort). This way anyone can view the results (PCs or not).

Thanks a lot!
Bob

Raw clips first though. That's untouched and original. Once we have them we can make tiff's and etc for everyone.

Sounds good to me.

Michael's link didn't copy over correctly:
Here is the working one:

Link to the chart:
http://www.graphics.cornell.edu/~wes...3-reschart.pdf

Bob

I don't have a rez chart in hand. Can someone provide me a file to print, if printing is good enough? I have a color laser printer and Canon inkjet here.

opps, there you go.

Thanks Robert for catching that!

Hi Kaku,

If possible, try to print the chart using black ink/toner only. This will give us the highest contrast. CMY (Cyan Magenta Yellow) combined alone gives only a muddish black color. Printing the chart on a photo-quality glossy paper will work the best if you have something like that around that you're willing to spare. Either way, after it's printed, you should be able to see black/white alterations in the 20 mark range (they're pretty small so you might need a magnifying glass ;-).

(insert) I printed this chart on 8.5x11in at 1200dpi, and it wasn't quite good enough. The 10 mark was the smallest line size that resolves clearly. 11x17in size paper will help if the printer rez can't get high enough for the smaller format.

Thanks a lot!
Bob

shooting a res target
 

When shooting the res target, try to use flat lighting to avoid paper gloss. The white part of the paper should not be blown out in the video, a light gray shade would be much better. Black shouldn't clip either. Motion blur is a factor, use a tripod. The lens will preform better at higher f stops. The zoom amount will also be a factor, I'd try several to see where the sweet spot is.
But for now, we would love to see anything! But people might remember the first image for a long time , gotta be careful !

Oh, turn off all the edge sharpening you can, that can always be added later with higher quality than the real time one in the camera.

Can't wait to see what it looks like!

-Les

Lighting res chart
 

Just to build on Les's suggestion: best way to light it is the way one shoots all flat artwork: 2 lights, one from either side at 45 deg angles. If you only have one light available, then you can get by with just one at a 45 deg angle. Great that you're doing all this!

If printing out a res chart, make sure you do it on a very high DPI printer, at least 1200 DPI. And frame the chart so that it fills the screen from arrow to arrow (charts usually have alignment arrows on them pointing to each edge) and get it as level as possible, with the camera as square to the chart as possible. And try to light it such that your iris is around f/4 to f/5.6, somewhere in that range.

I printed the chart on a mid grade printer, Epson c86, and it resolved the '10' resolution level with ease. I'm guessing that the HVX200 will be lucky to see details in the 8 level.
To any HVX200 users ( that are *allowed* to ) do this res test and post the result, I'd just make sure you can see the 10 level with a loupe before using the printout.
Perhaps we should invite others to test in the other manufacturers threads as well?
-Les

This is what I have to color/bw laser printing, so if I print the chart in A3 size, it must be okay then. I also have a B0 size Epson Maxart inkjet, so with that, we can even print it larger (then the lighting would harder). I'm equipped with a couple of decent lights by Lowel, so they should be good enough. I will make sure it gets lighted evenly.

Can't wait to see results!

It would be nice to know if it can beat Canon H1, which at 1080p recorded to tape resolves 800Hx650V lines and should have higher resolution via HDSDI.

The lens will be a factor in the resolution test. The lens resolving factor is dependent on iris opening (F-stop) and focal length. It would be nice to get frame grabs on the Panasonic at widest angle, longest telephoto, and couple shots in between these extremes, also at maximum iris opening and with the iris closed somewhat. At wide open iris and max. closed iris the resolution is likely to be lower.

Focus is extremely important. If you don't have high end HD monitor, let the camera auto focus, use focus assist, peaking, etc.

If at wide angle the chart does not fit the whole frame, use more than one chart; one in the middle, one a corner of the frame. The lens may not have a flat focus at wide angle so if the middle is in focus, the corner may not be, so two frame grabs with different focus settings may be needed. The results can be calculated from this setup even if the chart does not fit the frame properly.

Make sure the camera is on a tripod and use remote control.

The chrts need to be B/W and lit, hopefully softly, from sides so no reflections are created.

Is anybody paying Kaku for all this testing he's being asked to do? It's almost like he's got a part time job being the unofficial HVX-forum tester here!

Kaku - you should setup a Paypal account and take requests for various tests. At least you'd be compensated for all this work you're doing for us!!

(^_^)

Thanks for your concern, Robert.

As it is now, I could do this much and I'm doing the test for gainning my knowledge with the cooperation of the members here. I'm using the equipment belong to my company CreativeSuite (this name was
used and registered before Adobe started to use it for their products) and it is a world wide demonstration of what we can do for demonstrations, too.

I gained so much from here, so this is my and my company's contribution to this society. Every company that sells LCP and Plsma Panels have bought NLE systems from CreativeSuite. That explains the systems that we can build are flexible and reliable to use for testing of creating and handling different formats. So, please just remember that there is CreativeSuite in Japan that could do Mac and Windows based system for NLE, and on music side, Musetex that does the same for music. I hope foreign corporations that come to Japan can rely on us so ends up my company makes money.

If there was such a thing as a DV Info Net payroll, I'd put Kaku on it. He deserves it!

This is an all-volunteer thing though.

Hahaha, thanks Chris.

I'm spending much more time in video so our people in music division are jealous. I'm supposed to do more in music in this year, so I let my guys at CreativeSuite handle this test, which is very good experience and education for them to be associated with.

Kei Kinoshita will be the person.

Hi Kaku, what's the ETA on the res tests, just curious?


pappas

we probably finish doing this in a couple of days.

The sooner the better! As always thanks kaku!

I decided to send the chart data to output shop. It will be printed with camera ready quality. This should be something that noone would complain, am I right? If there is anyone with prepress knowlege, please help me with concrete output requirement for printing this chart like it should be 180 lines per inch and stuff.

By the way, my trusty staff at CreativeSuite agreed to work on this and the greenback test so that they can learn from the process.

We decided to go with 3000dpi A3 size printing on photo ready.

That should arrive tomorrow.

I did take some but I realized that Petr had a good post in suggesting doing this test.

I'm setting up everything in a larger room to do the extreme far through extreme close shots of the chart.

The chart printed with 3000dpi camera ready was very good. I had to use the magnifier to see the finest lines to see if the are all crisp and clear.

OK, I can't wait

i just did a test, not extreme close but at 57cm close up in formats that Michael requested. If it is Mac file, I can post it right away, but before it goes public, I want someone professional to look at them. Can anyone be on iChat/AIM?

did it go well? what was the max number in 1080p?

around 750 as Jan suggested?

I got little tied up. Setting is done so I will do this first thing in the morning.

I want as many people to help me on iChat while I do this. I will have my video on too. So please people in proper knowledge join me to archive this test on ichat about 9 hours from now.

My initial test only show the resolution around 600, not 700.

"My initial test only show the resolution around 600, not 700."

Oh-oh

WHAT????

Say it ain't so Kaku. That's all you got from the HVX200?

I dont beleive it. Man....don't trust those paper print-outs.... I keep tell' y'all!

- ShannonRawls.com

Kaku, just post some frame grabs and let us analyze it. I don't believe that the resolution is this low. You are probably not reading the chart correctly. You should have plenty of light and the chart should be costrasty; that means that white should be white or nearly white. The chart should be 16x9 and the arrows should touch the edge of the screen. If they don't, it's fine too. We'll factor it in. I've read plenty of resolution charts. It's easy to determine the resolution. If you want me to read the charts privately first, I can give you my email address.

Man, Shannon, are you tiresome... enjoy your canon and let us live

We are all excited to see the results when they come in, but we need to analyze them objectively and stay focused.

Consider the following:

1) We have no results yet:

• a) We have no authority to draw conclusions at this point.

2) If Panasonic reports around 750lp/ph:

• a) At what camera and lens settings?
• b) At what contrast ratio does this 750lp/ph occur?
  • i) Is it safe to assume 30% contrast if not stated (Answer: Maybe)
• c) Which chart has been used?
  • Does anyone have access to this chart so we can repeat same test Panasonic did?

3) We need to calculate the MTF (Modulation Transfer Function) so that we can conclude repeatable results both accurately and objectively:

• a) Anyone with a capable image editing program will be able to repeat this test.

4) We need to determine calculation errors:

• a) Are the results over or under estimates?
  • i) Can we tell by how much?
• b) What are the sources for errors?
  • i) We could repeat the tests after correcting errors to obtain more accurate results.

The MTF is not difficult to compute. If anyone is curious as to "What the heck is MTF?", see this link: http://www.normankoren.com/Tutorials/MTF.html
It is somewhat technical, as it should be, but it gives you a general and more specific idea that introduces this tool.

Cheers!
Bob

Actually all we need is for Kaku to focus it correctly, have it lighted properly and make sure the whites is white or near white. We are not comparing to any Panasonic tests, or veryfying Panasonic's tests. I've never seen the Panasonic produced charts. We are comparing to charts we have on H1, Z1, HD100, etc., which may not have bewen taken under ideal contrast condition either, or the chart does not fit the frame fully, but the information is sufficient to compensate for any defficiencies and draw conclusions.

Kaku is already going overboard by having the chart printed at 3,000 dpi, when 300 would have been more than enough. He will apparently test it at various focal lengths and F-stops, which is extremely important.

Hi Petr,

That will work for the most part, but if anyone is going to quantify any certain number of resolution, such as 750lp/ph (or even a range of numbers for that matter), then the analysis needs to be done. Lack of that, we should be able to eyeball the charts from various cameras to gain a general idea about resolution.

Bob

GO Green GO
 

Hello Kaku

when you perform the test, I especially would like to see the result with a 'pure green filter' fitted to the HVX200 and the colour turned down

this should tell us what the native resolution of the green CCD is, and consequently how much interlaced resolution is via pixel shift

thanks

If you want to see how much is the pixel shift contributing to resolution, IMHO you need to run the following test:

Turn down red and blue totally. Have the camera see green only.

I think that this is not important though. What is important is how the camera performs, not how it achieves its performance.

The camera has no facility to turn down blue and red totally/independently

Only reducing colour gain/saturation can achieve this.





The external green filter is important to combat the gain applied via pixel shift

The resolution of the green sensor will be of particular importance for those wishing to make HD productions with a Matrix/Amelie colour timing.

I'm back on the ichat.