PDA

View Full Version : Comparison GH1 & EX3 - frame grabs



crunchy
07-16-2010, 11:30 AM
Today I compared Sony EX3 and GH1 in a forest. Here are three grabs from a footage.

All EX3 shots are mildly over-exposed (compared to GH1) and contrast is relatively low. However, it seems that EX3 is slightly sharper. All shots were taken with the same shutter speeds and lens openings. However, on GH1 ISO values were between 400 and 800, while EX3 had 0dB gain (and brighter picture as well). Surprising result, at least for me. On the other side, EX3 footage is slightly noisy also at 0dB gain and it shows slight purple fringing. GH1 footage is muddy (especially Picture 3 and slightly Picture 1).

Since pictures are larger than allowed size, click links to see them.

Picture 1 (F5.6, 1/60s, GH1: ISO 400, EX3: 0dB)
ex3 (http://dsc.ijs.si/3d/files/pictures/fig01_ex3.jpg)
gh1 (http://dsc.ijs.si/3d/files/pictures/fig01_gh1.jpg)

Picture 2 (F5.6, 1/60s, GH1: ISO 400, EX3: 0dB)
ex3 (http://dsc.ijs.si/3d/files/pictures/fig02_ex3.jpg)
gh1 (http://dsc.ijs.si/3d/files/pictures/fig02_gh1.jpg)

Picture 3 (F5.6, 1/60s, GH1: ISO 800, EX3: 0dB)
ex3 (http://dsc.ijs.si/3d/files/pictures/fig03_ex3.jpg)
gh1 (http://dsc.ijs.si/3d/files/pictures/fig03_gh1.jpg)

Note that I made a mistake by taking 50i instead of 25p on EX3.
I have used MarekV "perfect project" settings (lower, but safer bit-rate).

Direct comparison (on the same picture) is given here:
Fig1 (http://dsc.ijs.si/3d/files/pictures/comp_gh1_ex3_01.jpg)
Fig2 (http://dsc.ijs.si/3d/files/pictures/comp_gh1_ex3_02.jpg)
Fig3 (http://dsc.ijs.si/3d/files/pictures/comp_gh1_ex3_03.jpg)

Ian-T
07-16-2010, 11:41 AM
I dunno....aside from brightness levels I think they look on par in terms of sharpness. The last pic of the GH-1 seems muddy in the middle area but sharp in the forefront. That might be a codec issue.

EDIT: Also, what I notice in the GH-1's image is in a well lit area it's very detailed and sharp but gets muddier in the shaded areas.

crunchy
07-16-2010, 12:35 PM
Exactly, they are in par in terms of sharpness in well lit areas. So, overall, EX3 sharpness is higher (when taking into account GH1's mud). On the other side GH1 is very good. If only there is some kind of AVCHD "button" which can increase the quality of lower-lit areas.
By the way, I have added direct comparisons in the first post.

John Caballero
07-16-2010, 12:39 PM
Ok.

Hunter Hampton
07-16-2010, 01:12 PM
Just what I thought- thanks!

HDkilledFILM.
07-16-2010, 01:15 PM
Is this with a standard gh1 or a hacked gh1?

crunchy
07-16-2010, 02:36 PM
Hacked (GH13), but not so high bitrate. Settings are from MarekV's The 'perfect' project.

cowpunk52
07-16-2010, 02:43 PM
Really interesting - I gotta admit, though, that i thought the EX3 would be sharper than it appears in your sample. Are you sure there isn't a backfocus issue?

Either way, thanks for posting! I'm gonna try to do the same thing against my XF300 and see how that pans out. I'll post results this weekend.

edit: I just noticed that you took 50i on the EX3. AFAIK, shooting interlaced on a progressive CMOS sensor will result in a loss of resolution, so maybe that's why the EX3 samples seem soft to me. This probably isn't a fair comparison between the two cams, I think the EX3 could perform much better.

HDkilledFILM.
07-17-2010, 11:13 AM
edit: I just noticed that you took 50i on the EX3. AFAIK, shooting interlaced on a progressive CMOS sensor will result in a loss of resolution, so maybe that's why the EX3 samples seem soft to me. This probably isn't a fair comparison between the two cams, I think the EX3 could perform much better.

I thought that as well... The images should also both be exposed correctly.

crunchy
07-17-2010, 11:28 AM
EX3 had switched odd ND filters, it was in manual aperture mode and shutter speed was controlled as well. GH1 did not have attached ND filter. During taking shots I read the values on both cameras, so there was no mistake. It's startling for me that EX3 was brighter than GH1 at the same opening and shutter speed even though EX3 gain was 0dB and GH1 was set at ISO 400 or 800. Can anybody explain it? This is very strange. It's true that there was some very subtle noise present on EX3 footage. I should increase brightness of GH1, but I did not want to add noticeable noise by increasing ISO too much.

Sensitivity of EX3 was the same in 50i and 25p mode (I checked it later on) for the same 1/60s shutter speed.

Barry_Green
07-17-2010, 12:49 PM
The EX1/EX3 are most definitely not the same sensitivity when you swap from 50i to 25p. They're a full stop more sensitive in 50i than they are in 25p. 400 ISO in 1080/25p, 800 ISO in 1080/50i.

crunchy
07-17-2010, 02:01 PM
Barry, what ISO are you talking about? AFAIK, EX3 has dB. On my tests it was fixed to 0dB.

Not the same sensitivity? This is strange. When I changed the modes I really noticed change in lens opening when it was in AUTO mode. However, later on I repeated the experiment 2 more times, but the lens opening remained the same. Still don't know what was the reason. Will have to repeat it as well.

In my opinion higher sensitivity in 50i mode might be due to interlaced/progressive reading of the sensor. In interlaced mode the line is read as sum of two neighbouring lines. However, it holds for CCD chips and I don't know if it holds for CMOS.

Anyhow, I'll have to repeat the experiment with EX3. Still, can you comment that brightness of EX3 is higher than GH1 at the same speed/opening and with 0dB vs. ISO 400-800?!?

dcloud
07-17-2010, 02:08 PM
wow. cant wait for af100 :D

Barry_Green
07-17-2010, 02:58 PM
Barry, what ISO are you talking about?
"rating" the camera as to its base sensitivity, it is approximately 400 ISO when in 1080p mode, 500 when in 720p mode, and 800 when in 1080i mode. Not talking about an adjustment, just about where it defaults at when at 0dB.


This is strange. When I changed the modes I really noticed change in lens opening when it was in AUTO mode. However, later on I repeated the experiment 2 more times, but the lens opening remained the same. Still don't know what was the reason. Will have to repeat it as well.
Might have been because you were at the maximum or minimum aperture already or something, but there's definitely a difference in the different modes. Also, be aware that the Sony lens iris readout doesn't tell you the whole story. The lens ramps from f/1.9 at wide angle to f/2.8 at tele, but the display never updates to tell you that, it still reports f/1.9 even though the actual iris is f/2.8.


In my opinion higher sensitivity in 50i mode might be due to interlaced/progressive reading of the sensor. In interlaced mode the line is read as sum of two neighbouring lines.
If that isn't exactly what it is, it acts an awful lot like it. :thumbsup:


Anyhow, I'll have to repeat the experiment with EX3. Still, can you comment that brightness of EX3 is higher than GH1 at the same speed/opening and with 0dB vs. ISO 400-800?!?
ISO is unfortunately not as standardized as we'd like it to be. It's oft reported that a Canon at 800 ISO responds like a GH1 at 400 ISO. You can't take any of those readings or reportings at face value, you have to test them in various circumstances to see what they really mean. Plus, it depends heavily on what gamma curve you're using in each. For example, using STD3 on the EX3 will result in a noticeably "brighter" picture than when using one of the CINE gammas.

crunchy
07-18-2010, 12:55 AM
"rating" the camera as to its base sensitivity, it is approximately 400 ISO when in 1080p mode, 500 when in 720p mode, and 800 when in 1080i mode. Not talking about an adjustment, just about where it defaults at when at 0dB.

Thank you very much on your valuable information! Where have you got this data?


The lens ramps from f/1.9 at wide angle to f/2.8 at tele, but the display never updates to tell you that, it still reports f/1.9 even though the actual iris is f/2.8.

Yes, that could be the reason. Thanks!


For example, using STD3 on the EX3 will result in a noticeably "brighter" picture than when using one of the CINE gammas.

Next time I will have to check other settings as well. Thank you again!

PappasArts
07-18-2010, 03:12 AM
If only there is some kind of AVCHD "button" which can increase the quality of lower-lit areas.
By the way, I have added direct comparisons in the first post.

Increase your bit rate- Since your doing 50i and not native- you can go double what your doing with out issue's. Use C ( 50-52-60 / 50i ) settings and you'll have zero mud problems.

BTW- cool tests. The EX performed as I expected it to do for a small sensor video camera competing with a way larger sensor densely packed in pixels. The GH1 looked fantastic.....

Pappas

crunchy
07-19-2010, 07:21 AM
Not easy to increase the bitrate so much. I might run into problems. Namely, I am using Transcend 16GB class 10 card. According to the following tread:

http://www.dvxuser.com/V6/showthread.php?t=217260

I measured writing speed at minimum of about 50Mb/s when using 32kB chunks (by the way, is GH1 using 32kB blocks?). Taking into account at least 20% reserve, it comes into about 40Mb/s. This is probably the highest reliable writing speed.

It would be useful if everybody would test his/her card and report results.

svecher
07-19-2010, 07:36 AM
Not easy to increase the bitrate so much. I might run into problems. Namely, I am using Transcend 16GB class 10 card. According to the following tread:

Get a Sandisk Extreme then. See? THAT WAS EASY (c) ;)

PappasArts
07-19-2010, 03:17 PM
Not easy to increase the bitrate so much. I might run into problems. Namely, I am using Transcend 16GB class 10 card. According to the following tread:

http://www.dvxuser.com/V6/showthread.php?t=217260

I measured writing speed at minimum of about 50Mb/s when using 32kB chunks (by the way, is GH1 using 32kB blocks?). Taking into account at least 20% reserve, it comes into about 40Mb/s. This is probably the highest reliable writing speed.

It would be useful if everybody would test his/her card and report results.


"C" settings and 24/25fps over 50/60i have zero issues.... And that is double the bit rate ceiling of what your using. It doesn't makes sense not to use those settings.

Pappas

crunchy
07-20-2010, 01:42 AM
Increase your bit rate- Since your doing 50i and not native- you can go double what your doing with out issue's. Use C ( 50-52-60 / 50i ) settings and you'll have zero mud problems.

Are you sure that you get zero mud problems in lower-exposed areas (when you have mixed normal-lit and low-lit areas)? My experience was that I still get some mud even when I used C settings.