September 23, 2014, 10:41:22 PM

Show Posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.


Messages - heptagon

Pages: [1] 2 3 ... 11
1
EOS Bodies / Re: Sample Images From the EOS 7D Mark II
« on: September 21, 2014, 03:58:21 AM »
Color and pattern noise give us an idea about sensor performance, but they are not taken into consideration when calculating DR. So you should leave Noise Ninja checked on at default setting too (chroma NR 50) and look at the luminance noise. Make sure that Detail slider is set to zero as it tends to enhance noise. Now throw in D7100 (DR 13.9 according to DXO) in the mix and see what happens. You won't see much difference in noise. So, I don't know if we can take this test too seriously to get an idea about DR. We need proper shadow lifting and read noise measurement. Other caveat is that 7DII is not supported officially by any RAW converter yet, we can't really rely on these tests. 
Something else I have noticed and not sure what to make of this. Take this 3 cameras, do above adjustments, but leave the shadow slider at zero. Export the JPEG files to LR or CS and adjust exposure by +1 (or save as JPEGs and open again with PN and adjust exposure by +1). Now look at the cup, it is darker in 7DII file compared to other two cameras. Look at the whole scene, everywhere 7DII retains more color and detail (except for overblown areas) compared to 7D and D7100. Is it because 7DII scene is slightly underexposed? Lighting has changed? RAW converter needs fine tuning? Something else?

Ok, I agree, lifting the shadows might not do what i expected it to do. For a proper comparison we should use other programs where whe know better what happens. Also the scene here doesn't contain the amount of dynamic range we need for a proper comparison. The bright shadows of the 7D are weird, maybe the program has a wrong offset for black... don't know.

What we can say for sure though, is that the 7DII has better shadow noise than the 7D.

2
EOS Bodies / Re: Sample Images From the EOS 7D Mark II
« on: September 20, 2014, 12:08:01 PM »
Test images:

7D2:
http://www.imaging-resource.com/PRODS/canon-7d-mark-ii/E7D2hSLI00100NR0.CR2.HTM
7D:
http://www.imaging-resource.com/PRODS/E7D/E7DhSLI00100_NR_0.CR2.HTM

Program: Photo Ninja
http://www.picturecode.com/download.php

Settings:
Only the first three checkmarks on:
* Demosaic
* Color correction
* Exposure and detail
(no extra sharpness/color/noise correction)

Everything on default except in Exposure and Detail:
* Exposure offset +3
* Shadows +1

Now you see the colored noise in the cloth and cup at the right bottom.

3
EOS Bodies / Re: Sample Images From the EOS 7D Mark II
« on: September 20, 2014, 02:09:30 AM »
I opened the CR2 files at 100ISO of the 7D and 7DII in PhotoNinja, turned off all noise/color enhancements and pushed the shadows:
* Pattern noise seems to be gone.
* Shadows are still noisy. Maybe 1/3rd Stop or 1/2 Stop more dynamic range than the 7D. But more useable due to the lower Pattern noise.

So, it is the same sensor technology with little enhancements.

4
EOS Bodies / Re: A New EOS Pro Body With 46mp Next Month? [CR1]
« on: September 18, 2014, 06:34:37 AM »
Reply to the 30MPix nonsense.

Current sensors have a bayer pattern providing with only 1/4th resolution in red/blue. So make that makes the 30MPix into 120MPix if you want a color sensor. But it doesn't stop there. Look at the resolution of the 300mm 2.8 IS II. It outperforms current sensors when using the 2x extender on crop. That'd be 18 MPixels times 4 (extender) times 2.56 (crop to FF area) resulting in 184 MPix. This times 4 because of the bayer pattern is 737 MPix for full color resolution.

Even though most lenses will only provide this resolution in the center of the image at the optimum aperture (<<f/8) we're far from "too many pixels on this sensor".

Admittedly fewer and fewer people will benefit from more pixels in fewer and fewer situations since you need super fast shutter speeds, perfect focus, optimum lighting etc. but if technology provides more pixels at moderate cost, why not take it to get those 10% of the shots you couldn't with the old tech?

In the end the real loser will be the 1.4x and 2x extenders which are just a magnifying glass and crop cameras for tele which just use a smaller imaging circle.

5
This simply means that Canon and maybe many of the other Camera manufacturers have not done their (law/patent) research properly and now need to pay for a license and a fine to use technology others have invented and patented in the USA (and probably other countries).

This probably makes DSLRs more expensive but in the end is no big deal.

6
EOS Bodies / Re: A New EOS Pro Body With 46mp Next Month? [CR1]
« on: September 18, 2014, 05:14:54 AM »
A quick 2 questions: 

1) Has anyone asked Canon directly why they don't offer a higher-mpixel camera?  Asking someone other than a marketing guy, that is.

2) Has anyone from Canon provided a clear, reasonable reply to why their current product offering is as good, if not better than Nikon/Sony at 36mpixel (rumored to soon go to 46mpixel)?

A link to something (Presentation?  Whitepaper? Engineering paper?) would be great.  I'd like to take the time to understand (as best I can) why Canon is "stuck" in it's current position.

1) Only public relation persons are allowed to speak to the public. If an engineer answers he'll lose his job and possibly get a big lawsuit.

2) It still sells well enough, so it must be good enough. Looking at the sensor alone, Canon is way behind in some important areas. In my opinion they should adapt a Sony sensor for a high density studio/landscaping body. This would make many people happy.

7
EOS Bodies / Re: How excited are you about the new 7D II?
« on: September 16, 2014, 03:25:59 AM »
I guess it depends on your purpose of purchase. As a photographer, I can see the idea of this camera, although…. why not either boost the sensitivity of the chip to compete with Sony's A7s or boost the pixels so you don't feel like you are licking the 5D3 in the ass….

What could really make the Canon 7D II look like bullshit is the Samsung NX1 which was just announced:

http://www.nofilmschool.com/2014/09/samsung-nx1-new-4k-mirrorless-shoots-h265

Quote
The specs:

28.2MP BSI APS-C CMOS sensor
4096 x 2160: 24 fps
3840 x 2160: 30 fps
1920 x 1080: 60 fps, 30 fps, 24 fps, 50 fps, 25 fps
1280 x 720: 60 fps, 30 fps, 24 fps, 50 fps, 25 fps
640 x 480: 60 fps, 30 fps, 24 fps, 50 fps, 25 fps
Clean HDMI 4:2:2 (8-bit)
H.265 Codec applied to 4K and UHD video files
Tiltable 3.0″ HD 1036k-dot touchscreen
205 phase-detection AF points
15 fps burst mode
1/8000 Max Shutter Speed
ISO range from 100 to 51,200
Camera can utilize UHS-I or UHS-II-SD-type memory cards
1/8" Headphone, 1/8" Microphone, HDMI D (Micro), USB 3.0
802.11b/g/n/ac Wi-Fi capability coupled with Bluetooth 3.0 technology
Samsung NX Lens Mount
Price: $1,500 Body Only, $2,800 with 16-50mm f/2-2.8 S ED OIS Lens

Unfortunately the NX mount is made incompatible to the EF mount.

While the NX1 has better specs and the better sensor it lacks one thing and that is the advanced phase AF of the 7DII. If Samsung made an NX1 with an fully functional EF mount it would win hands down in all categories except maybe low light sports.

8
EOS Bodies / Re: Video Review: Canon EOS 7D Mark II
« on: September 15, 2014, 01:18:16 PM »
Review? More like professional commercial spot.

9
Is this relevant for focusing speed? (E.g. Supertele+Extender @ f/8)

If no, then it's pretty irrelevant for the target audience because the sensor is too small for low light action shots.

10
EOS Bodies / Re: Canon EOS 7D Mark II Specifications Confirmed
« on: September 12, 2014, 11:12:28 AM »
This is how I shoot low light on my 30D..  I stop at ISO800..   here's the implied readout noise levels from the DxO data...  on the D800 looks like ISO200 is worth having but benefit rapidly drops beyond that.

Coincidentally the D810 now has native ISO64 because to improve DR and SNR.

For studio and landscape cameras the way to go is more Megapixels and lower ISOs.

11
EOS Bodies / Re: Canon EOS 7D Mark II Specifications Confirmed
« on: September 12, 2014, 10:06:51 AM »
If you shoot raw, then stick with ISO100 and be prepared to "underexpose" then fix in post.


well that's not good advice! :D

Care to explain why shooting at ISO100 but underexposed by 2 stops is worse than shooting at ISO400 on a Nikon 800/600 series?

Numbers are useful here.


You have two main sources of noise at low ISO:
1) Quantum noise
2) Readout noise

Quantum noise is due to the fact that photons behave like particles and if you on average expect 10000 photons in a pixel in reality you get sometimes more and sometimes less leading to a Poisson distribution around that average value. The quatum noise is the square root of the number of photons e.g. 100 for 10000 photons and 10 for 100 photons.

The readout noise is various electronic noise and depends on the amplification factor and other interference. With the new Sony sensors the readout noise is virtually the same on ISO 100 to ISO 6400 e.g. 5 electrons. With the canon sensors the readout noise is better for high ISO e.g. 3 electrons but worse at low ISO e.g. 30 electrons at ISO 100.

If you look at a bright pixel you get 10000 electrons from the photons +-100 electrons due to Poisson statistics (even a perfect sensor will get that). The S/N ratio is 100.

If you look at a dark pixel you get 100 electrons from photons and +-10 electrons from Poisson statistics. The S/N ratio is 10.

Now factor in the read noise.
- Bright pixel:
-- Canon: 100+30, S/N 77
-- Nikon: 100+5, S/N 95
- Not much difference.

- Dark pixel:
-- Canon: 10+30, S/N 2.5
-- Nikon: 10+5, S/N 6.7
- Almost a factor of 3!

If you do the same calculation at high ISO, the Canon sensor gets a little advantage.


From this we can conclude that BRIGHT pixels have the SAME QUALITY with a Canon and a Sony sensor.
But DARK pixels suffer from readout noise and here the Sony sensor is much better.

So if you do not underexpose or lift the shadows in your pictures, you will be OK with current Canon sensors.


Oh boy....


I do this stuff every day for work.

Yes poisson noise is also present, but as the two situations I'm comparing here have the same light levels, the poissn noise is identical, so we can drop it out of the equations.

Also you don't just add the noise.   it's an RMS..  (in volts) so that changes the maths a little,  Sqrt(Noise A+ Noise B)

Yes, I should have used RMS addition of the independent noise.

No, the Poisson noise is important to calculate the S/N value which considers all noise sources.


After all, this exactly explains why cameras with the same sensor size have the same SNR18% performance (in print mode) on DXO. Also it explains why the dynamic range is worse on Canon sensors compared to new Nikon/Sony sensors. Also it explains why Canon has equal or better high ISO performance.

12
EOS Bodies / Re: Canon EOS 7D Mark II Specifications Confirmed
« on: September 12, 2014, 09:35:12 AM »
If you shoot raw, then stick with ISO100 and be prepared to "underexpose" then fix in post.


well that's not good advice! :D

Care to explain why shooting at ISO100 but underexposed by 2 stops is worse than shooting at ISO400 on a Nikon 800/600 series?

Numbers are useful here.


You have two main sources of noise at low ISO:
1) Quantum noise
2) Readout noise

Quantum noise is due to the fact that photons behave like particles and if you on average expect 10000 photons in a pixel in reality you get sometimes more and sometimes less leading to a Poisson distribution around that average value. The quatum noise is the square root of the number of photons e.g. 100 for 10000 photons and 10 for 100 photons.

The readout noise is various electronic noise and depends on the amplification factor and other interference. With the new Sony sensors the readout noise is virtually the same on ISO 100 to ISO 6400 e.g. 5 electrons. With the canon sensors the readout noise is better for high ISO e.g. 3 electrons but worse at low ISO e.g. 30 electrons at ISO 100.

If you look at a bright pixel you get 10000 electrons from the photons +-100 electrons due to Poisson statistics (even a perfect sensor will get that). The S/N ratio is 100.

If you look at a dark pixel you get 100 electrons from photons and +-10 electrons from Poisson statistics. The S/N ratio is 10.

Now factor in the read noise.
- Bright pixel:
-- Canon: 100 and 30 => 104 total, S/N 96
-- Nikon: 100 and 5 => 100 total, S/N 100
- Not much difference.

- Dark pixel:
-- Canon: 10 and 30 => 32 total, S/N 3.1
-- Nikon: 10 and 5 =>11 total, S/N 9.1
- Almost a factor of 3!

If you do the same calculation at high ISO, the Canon sensor gets a little advantage.


From this we can conclude that BRIGHT pixels have the SAME QUALITY with a Canon and a Sony sensor.
But DARK pixels suffer from readout noise and here the Sony sensor is much better.

So if you do not underexpose or lift the shadows in your pictures, you will be OK with current Canon sensors.


EDIT: Use RMS for noise addition.

13
Third Party Manufacturers / Re: Zeiss Otus 85mm f/1.4 Reviews
« on: September 12, 2014, 08:00:19 AM »
DxOMark shows that this lens outresolves any of the current Canon fullframe sensors. It's been a long time since the last 1Ds...

14
EOS Bodies / Re: A Rundown of Canon at Photokina
« on: September 04, 2014, 10:36:29 AM »
4k for recording is not the same as 4k for playback.

Just compare mushy compressed 1080p with RAW 1080p. Having the option of using 4k for recording, then cropping (e.g. stabilization) and downscaling the compressed version of the video stream would be a very sensible option in many cases to produce high quality 1080p. So, no, it is not only a gimmick for people who know what they do with video.

15
EOS Bodies / Re: Plan B
« on: August 07, 2014, 07:26:13 AM »
i don´t care much about aps-c.

im only interested in the 7D MK2 because i want to see what canon does and if there will be an improvement in sensor technology.

Seeing the 7D mainly as a tele/action/sports camera probably something along the lines of a good phase detection AF + improved DualPixel technology. Maybe QuadPixels. Possibly initial focus acquired by the dedicated AF sensor, then handing off to Sensor focus to follow the objects and shoot at a high frame rate without moving the mirror. This would require a hybrid viewfinder.

So for sensor technology this doesn't mean necessarily mean very much improvement in regards to quality of low ISO still images. I wouldn't expect too much in this regard.

Pages: [1] 2 3 ... 11