GH4 Andrew Reid: Is this the new GH5 Sensor?

P

Peter Berry

Well-known member
http://www.eoshd.com/2016/11/olympus-e-m1-ii-hands-on-mini-review-a-preview-of-the-panasonic-gh5/

Quoted:

"I got to play with the new Olympus E-M1 II today, a pro Micro Four Thirds camera they hope will sell for $2000. It certainly pulls out all the stops to imitate a miniature 1D X Mark II.

This camera does NOT crop the sensor in 4K mode like the GH4. Instead, it over-samples from 5K down to 4K, using the full width of the sensor even with sensor-shift 5 axis stabilisation enabled.

Featuring a 237Mbit codec, the image quality in Cinema 4K mode is outstanding with very little rolling shutter.

I am wondering if the E-M1 II is a preview of the fundamental technology (sensor, codec, image processing) that is going to be in the Panasonic GH5 when it arrives in 2017.

It’s certainly higher specced for video than I thought it would be for an Olympus camera.

The all-new 20MP sensor does a full pixel readout in Cinema 4K mode which is a big advantage already over the current GH4, and it should be because it’s nearly 3 years newer.

Even with the 5 axis image stabilisation enabled, there’s no extra crop. The only time a crop of the sensor happens is when added electronic image stabilisation is enabled on top of the already very capable 5 axis sensor based stabilisation. Except in emergency shot-grabbing situations, I wouldn’t turn it on. It produces some tearing and distortion too."


Add the supposed "knowns" of internal 10-bit 4:2:2, 5-axis IBIS, 60p 4K, and 30p "6K Photo" (18.7MB frames in 3:2 or 4:3 with an insignificant crop), and you have the 5th gen. of ground-breaking GH bodies...

Peter
 
I read a recent review of the Oly E-M1-II 4K mode saying that UHD files were significantly softer than its C-4K, suggesting less clean down-sampling of the full sensor. No mention of moire, though. I imagine that Panny will offer a central sensor 1:1 readout as a mild 4K ETC mode, with a 2.7X crop factor for UHD and about 2.5 for C-4K based on the Oly's 5184x3888 pixels.

I do find my 12-35/2.8 coming up short pretty often in single cam stage productions, and a tap of the ETC button to a 47mm top end would be mighty slick.
 
The thing is, this not being a full frame sensor, you are looking at a more limited dynamic range to start with. I don’t believe DXOMark, I believe me own eyes more and I think the larger the sensor, the better scenes requiring high dynamic range always look. The E-M1 II didn’t show much improvement over the old model in terms of high ISO performance either.

Sure with raw stills you can expose for the highlights and shove the shadows up into the mid-tones for an awful looking image DPReview will claim is 13 stops… whatever.
Click to expand...

Darn.

The resolution and other specs sound great though. Would have been thrilled to see DR to compete with the BM m43 camera, and better lowlight/high ISO performance.
 
It's the same old Sony sensor. There are apparently a lot of similarities between Olympus and Panasonic but the end results could be slightly different.
 
DLD said:
It's the same old Sony sensor. There are apparently a lot of similarities between Olympus and Panasonic but the end results could be slightly different.
Click to expand...

Which "same old Sony sensor" are you referring to? Panasonic says it's a new one, and AFAIK the last year's Pan. GX8 and Oly Pen-F have been the only m4/3 bodies with a 20MP sensor.
 
Peter Berry said:
Which "same old Sony sensor" are you referring to? Panasonic says it's a new one, and AFAIK the last year's Pan. GX8 and Oly Pen-F have been the only m4/3 bodies with a 20MP sensor.
Click to expand...
That's the one (also in DVX200). Sony, nor anyone else, doesn't have any other MFT's with 20 MPX.
 
Current sensor quality is quite good. Faster reading and good down sampling and processing can improve video quality very much. Think GH4 10bit 4.6k RAW photo down sampled to 3840x2160 with minimal rolling shutter. The quality would be almost perfect.
 
The sensor also doesn't do all of the work when it comes to the image. You also have to factor in the DSP or the processor in the camera. All sensors pump out raw data and it is up to the processors in the camera to take that raw signal and debayer it, scale it, apply picture profiles, white balance and everything else that turns a raw set of data into a usable image that we love to look at. The sensor just makes the raw image data from light hitting it.

You can have two cameras use the same sensor that deliver very different results due to how well that data is processed on the fly to produce multiple frames per second. Things like better noise reduction can make the world of difference at higher ISO levels and give the illusion that a particular camera is better in low light when in fact that raw data coming off the sensor may be just as noisy. A better processor also means a proper image scale vs pixel binning which can also make a huge difference to the image quality. The sensor doesn't do the scaling. The processors do that.

Think of the sensor as a bucket of legos you hand to two kids. Each kid has the same nice legos to build something with but one kid may have more creative talent than the other and build something nicer looking. Of course how good that bucket of legos is will also affect the outcome of what the kids build but the kids also play a part in the quality that comes out of those buckets of legos.
 
Thomas Smet said:
The sensor also doesn't do all of the work when it comes to the image. You also have to factor in the DSP or the processor in the camera. All sensors pump out raw data and it is up to the processors in the camera to take that raw signal and debayer it, scale it, apply picture profiles, white balance and everything else that turns a raw set of data into a usable image that we love to look at. The sensor just makes the raw image data from light hitting it.

You can have two cameras use the same sensor that deliver very different results due to how well that data is processed on the fly to produce multiple frames per second. Things like better noise reduction can make the world of difference at higher ISO levels and give the illusion that a particular camera is better in low light when in fact that raw data coming off the sensor may be just as noisy. A better processor also means a proper image scale vs pixel binning which can also make a huge difference to the image quality. The sensor doesn't do the scaling. The processors do that.

Think of the sensor as a bucket of legos you hand to two kids. Each kid has the same nice legos to build something with but one kid may have more creative talent than the other and build something nicer looking. Of course how good that bucket of legos is will also affect the outcome of what the kids build but the kids also play a part in the quality that comes out of those buckets of legos.
Click to expand...

Brilliantly put. Thanks for saying it!
 
DLD said:
That's the one (also in DVX200). Sony, nor anyone else, doesn't have any other MFT's with 20 MPX.
Click to expand...

The DVX 200?? Why would a pro m4/3 camcorder use a 20MP sensor when it's max. res. is 4096x2160, or 8.8MP allowing nice fat pixels for better hi-iso performance? Nothing in its specs I could find confirming a downsampled much larger sensor.
 
Thomas Smet said:
The sensor also doesn't do all of the work when it comes to the image. You also have to factor in the DSP or the processor in the camera. All sensors pump out raw data and it is up to the processors in the camera to take that raw signal and debayer it, scale it, apply picture profiles, white balance and everything else that turns a raw set of data into a usable image that we love to look at. The sensor just makes the raw image data from light hitting it.

You can have two cameras use the same sensor that deliver very different results due to how well that data is processed on the fly to produce multiple frames per second. Things like better noise reduction can make the world of difference at higher ISO levels and give the illusion that a particular camera is better in low light when in fact that raw data coming off the sensor may be just as noisy. A better processor also means a proper image scale vs pixel binning which can also make a huge difference to the image quality. The sensor doesn't do the scaling. The processors do that.

Think of the sensor as a bucket of legos you hand to two kids. Each kid has the same nice legos to build something with but one kid may have more creative talent than the other and build something nicer looking. Of course how good that bucket of legos is will also affect the outcome of what the kids build but the kids also play a part in the quality that comes out of those buckets of legos.
Click to expand...

So true, and the announced new GH5 capabilities of internal 10-bit 4:2:2, UHD 60p, scaling, and concurrent 5-axis IBIS all point to a huge increase in sensor readout/processing speed and data throughput - which seems reflected in the Oly E-M1 II minimal 4K rolling shutter skew and 237Mbps MOV codec as reported by Andrew Reid.
 
Peter Berry said:
The DVX 200?? Why would a pro m4/3 camcorder use a 20MP sensor when it's max. res. is 4096x2160, or 8.8MP allowing nice fat pixels for better hi-iso performance? Nothing in its specs I could find confirming a downsampled much larger sensor.
Click to expand...
The DVX200 does use about 13 megapixels in 4K/UHD and scales that down to a 4K image. The results are extremely sharp 4K/UHD.

The new UX180 isn't an m43 sensor, it's a 1" sensor, but it does use native-sized pixels. As such, it about matches the DVX200 in sensitivity, even though the sensor is literally half the size.
 
Shenan said:
This rumor from July had mentioned that the image improvements on the GH5 would be coming from a "complete new" , "very powerful processor engine" with basically the same sensor from GX80/85: http://www.43rumors.com/ft5-panasonic-gh5/
Click to expand...

Very old hat, as since then at PK-16 Panasonic announced a 20MP sensor and "Photo 6K" stills in 3:2 and 4:3 format of 18.7MP. I suppose it could be a development of the 20MP GX8 sensor....
 
Peter Berry said:
Very old hat, as since then at PK-16 Panasonic announced a 20MP sensor and "Photo 6K" stills in 3:2 and 4:3 format of 18.7MP. I suppose it could be a development of the 20MP GX8 sensor....
Click to expand...

Ah, yes, I had forgotten those details, thanks! The interesting part of that rumor was how much they seemed to be emphasizing processor power. Let's see what they do.
 
Peter Berry said:
The DVX 200?? Why would a pro m4/3 camcorder use a 20MP sensor when it's max. res. is 4096x2160, or 8.8MP allowing nice fat pixels for better hi-iso performance? Nothing in its specs I could find confirming a downsampled much larger sensor.
Click to expand...
GH5 will (allegedly but almost certainly) have a full sensor/6K readout. DVX200, as Barry had mentioned, uses 13 MPX. GX8 uses the center-crop 8.8 MPX for 4K.
 
DLD said:
GH5 will (allegedly but almost certainly) have a full sensor/6K readout. DVX200, as Barry had mentioned, uses 13 MPX. GX8 uses the center-crop 8.8 MPX for 4K.
Click to expand...

The DVX200 4K clips I've seen are impressively sharp and hold up to 200% viewing, but the penalty for the scaling may be the reported somewhat more moire than expected. The GH5 will have many more pixels to play with in the full sensor readout, for whatever that's worth. Both Sony with it's 24MP sensor and the 28MP NX1 do a great job with their full sensor scaling RE sharpness. Rolling shutter - not so much. Andrew's report of the coming E-M1-II's minimal 4K skew with it's 20MP full sensor readout is encouraging for the GH5, which likely has even greater processing moxie.
 
You must log in or register to reply here.
Back
Top