6K photo, can't touch this with FF.

Tommi K1 wrote:

JaKing wrote:

Tommi, please read my post again (carefully), then read your reply. Then get back to me ... .

1.5 vs 256

Like to supply some reference/s for that?

I would just love to have some 256 GBps SSDs ...

Richandhiscat wrote:

tt321 wrote:

kolyy wrote:

Richandhiscat wrote:

Are you sure you read it ?

The title says it all "" 6K photo, can't touch this with FF. " so right from the get go the post was the sadly all too common FF target. I do not understand why they do not target the much closer in size and image quality 1" sensor cameras for their comparison . Instead they ignore that format jump over APS which in the case of Sony, Nikon, Fuji, Pentax all have superior image quality { especially base ISO DR } . Just to get to FF I don't think those MF guys should get off so lightly either.

You need to read it literally. No FF camera has a 6K photo mode at the moment, so they can't touch it (maybe should have said "do" instead of "touch"). And as he's using a full frame lens, then FF is the format to naturally compare it to.

You don't need a 6K Photo mode to do this. All that is needed is a 1:1 video read out from the center of the sensor. Panasonic has it for ages in the form of the ETC mode ("electronic teleconverter"). As FHD is enough due to the heavy crop, even my lowly GM5 could do it.

Sony has a similar feature for called "Clear Image Zoom". It's even more advanced as it does downsampling from an arbitrarily large area of the sensor. Thus with an A7R III, for example, one could use Clear Image Zoom in APS-C crop mode (which does full read out from the area) and just shoot regular video. There will be a difference due to pixel density, of course, but it's not that much as people here seem to think - the pixels of the GH5 are roughly 3/4 of the size of the ones in the A7R III (in linear terms).

In theory, this is not all true (I've not done anything like this so cannot comment about the practical implications of the theory).

What you are suggesting is to grab frames from video which is default compressed both spatially and temporally, which is originally intended for playing back as video. The 6K (or 4K in some cameras) photo mode is intended to produce individual still photo frames, to be viewed and used as single photographs, which hopefully contains zero temporal compression.

This might not be important if you are shooting the moon alone, as you have all sorts of algorithms with which you can construct a good single frame out of multiple frames from a temporally compressed video sequence. However, in this case, every single frame counts and you cannot construct a good ISS image out of a sequence of multiple images. And even if you find a single image where the ISS is good, its position might not be for your liking in terms of composition.

With a 6K photo sequence, you can pick any image you like, and you can even reconstruct the whole trajectory with multiple copies of the ISS over the moon in pp and every one of them would be good. Not that easy from a native video sequence.

Without looking it up, isn't 6K jpg compression as opposed to h264/5 video compression?

No, 6K is h265, 4K is h264. Both IPB (interframe) compression. In other words it's regular video with the usual compression artifacts.

tt321 wrote:

kolyy wrote:

Richandhiscat wrote:

Are you sure you read it ?

The title says it all "" 6K photo, can't touch this with FF. " so right from the get go the post was the sadly all too common FF target. I do not understand why they do not target the much closer in size and image quality 1" sensor cameras for their comparison . Instead they ignore that format jump over APS which in the case of Sony, Nikon, Fuji, Pentax all have superior image quality { especially base ISO DR } . Just to get to FF I don't think those MF guys should get off so lightly either.

You need to read it literally. No FF camera has a 6K photo mode at the moment, so they can't touch it (maybe should have said "do" instead of "touch"). And as he's using a full frame lens, then FF is the format to naturally compare it to.

You don't need a 6K Photo mode to do this. All that is needed is a 1:1 video read out from the center of the sensor. Panasonic has it for ages in the form of the ETC mode ("electronic teleconverter"). As FHD is enough due to the heavy crop, even my lowly GM5 could do it.

Sony has a similar feature for called "Clear Image Zoom". It's even more advanced as it does downsampling from an arbitrarily large area of the sensor. Thus with an A7R III, for example, one could use Clear Image Zoom in APS-C crop mode (which does full read out from the area) and just shoot regular video. There will be a difference due to pixel density, of course, but it's not that much as people here seem to think - the pixels of the GH5 are roughly 3/4 of the size of the ones in the A7R III (in linear terms).

In theory, this is not all true (I've not done anything like this so cannot comment about the practical implications of the theory).

What you are suggesting is to grab frames from video which is default compressed both spatially and temporally, which is originally intended for playing back as video. The 6K (or 4K in some cameras) photo mode is intended to produce individual still photo frames, to be viewed and used as single photographs, which hopefully contains zero temporal compression.

This might not be important if you are shooting the moon alone, as you have all sorts of algorithms with which you can construct a good single frame out of multiple frames from a temporally compressed video sequence. However, in this case, every single frame counts and you cannot construct a good ISS image out of a sequence of multiple images. And even if you find a single image where the ISS is good, its position might not be for your liking in terms of composition.

With a 6K photo sequence, you can pick any image you like, and you can even reconstruct the whole trajectory with multiple copies of the ISS over the moon in pp and every one of them would be good. Not that easy from a native video sequence.

You seem to be confused about how interframe compression works. All the frames you see are already reconstructed from temporal data.

JaKing wrote:

Tommi K1 wrote:

JaKing wrote:

Tommi, please read my post again (carefully), then read your reply. Then get back to me ... .

1.5 vs 256

Like to supply some reference/s for that?

I would just love to have some 256 GBps SSDs ...

You seem to be impressed by very mundane things. A single lane of PCI Express 3.0, the current most widely used standard in motherboards, is 1GB/s.  And you have dozens of those in a pc (16x on a cpu and some more on the chipset). And the bus from the CPU to RAM is an order of magnitude faster (typically 2x64bit at up to 3.2GHz with DDR4).

And you should compare SSDs with the SD cards in the camera.

So no, I am sorry, your favorite camera is not a supercomputer.

bobn2 wrote:

Tommi K1 wrote:

JaKing wrote:
Look at the processing grunt built in to my E-M1 MkII - 2x quad core CPUs, and they are putting data along the internal data bus at something like 1.3-1.5 *GB*ps! Struth, that's much faster than the internal data bus in any of our computers here ...

Nor really so.

The impressive thing is capability to process 60 raw files per second and convert those to JPEG.
Try that with any Intel or AMD processor...

Not too bad, since mostly those processors have on-chip GPUs these days. They tend to get ignored in top-end computers in favour of bigger, better GPUs on the PCIe, but they are quite up to demosaicking, mapping and JPEG encoding 60 files per second used properly. JPEG encoding is pretty trivial compared to things like H265.

Well, and then no camera can process 60 raw files per second and convert them to jpegs. It can just store them in the buffer and process later.

No one said it was ...

You must be very young not to be impressed with what is now achievable by a portable, battery powered device ... AND, I'm not just talking about a camera I happen to be fortunate enough to own, but the entirety of modern technological achievements, amongst the massive advances in most fields of human endeavour.

Sequencing the (a ... ) human genome took over twenty years the first time around, and cost billions (? - I don't actually know what the cost of the Human Genome Project was). These days, this process is a commonplace, and can be done in hours/days, not decades.

And also just BTW, I was talking about the data bus in OUR computers (not yours, or someone else's ... ) , for those who cannot read - also a commonplace these days, it appears.

JaKing wrote:

No one said it was ...

You must be very young not to be impressed with what is now achievable by a portable, battery powered device ... AND, I'm not just talking about a camera I happen to be fortunate enough to own, but the entirety of modern technological achievements, amongst the massive advances in most fields of human endeavour.

Sequencing the (a ... ) human genome took over twenty years the first time around, and cost billions (? - I don't actually know what the cost of the Human Genome Project was). These days, this process is a commonplace, and can be done in hours/days, not decades.

Oh, don't misunderstand me, I am pretty enthusiastic about the technological progress!

And also just BTW, I was talking about the data bus in OUR computers (not yours, or someone else's ... ) , for those who cannot read - also a commonplace these days, it appears.

I have to doubt that, unless you are running on computers from the previous millennium. Even the very first Pentium 4, introduced in 2000, had a front side bus capable of 3.2GB/s (4x100MHz, 64bit). The FSB of the Athlon XP was similarly capable. And there are not too many P4s and AXPs surviving in the wild today.