Canon 5D Mark IV -- Time to Discuss!

TeamSpeed wrote in post #18113528
2 stops underexposure on the 5D4, as long as you are at least 2 stops lower than the native ceiling of the camera should no longer produce unwarranted noise and poor results like in the past. This is the first camera that seems to be almost ISO-invariant as they call it.

A shot at ISO 400 underexposed by 4 stops and brought back up after the fact digitally looks almost identical to an actual ISO 6400 shot for example. That is something we have not been able to do with the old sensor design. The next round of changes to this new sensor should produce some nice returns for the consumer in this area!

The exposure triangle will become a dinosaur!

More exposure = more SNR; it should be as simple as that, and that's what it really is when you strip away all the quirks of primitive technology. ISO would only mean what level of exposure caused by something in the scene, rendered as middle gray in the display, receives. That doesn't even need to be decided at exposure time; you might make two different crops from the same exposure and render them at different ISOs, with no penalty other than more noise from less light.

BigAl007 wrote in post #18113594
Actually it's line pairs, not lines, so you have two rows of pixels at one ISO, and a second pair of rows for the other ISO value. It has to be row pairs, because if you used alternating rows, you would destroy your colour information. Unfortunately because the Bayer CFA uses alternating rows of Red/Green and Green/Blue filter pairs, the full colour data is only sampled at half the sensel resolution. Because you would normally be able to effectively look at both the RGGB and the GBRG combinations you do lose colour fidelity where you have horizontally running fine detail where you have the higher ISO rows blown. It is a shame that you cannot read out the sensor multiple times, to do true DualISO recording. I would love that, even if it meant a halving of the frame rate, since I am often shooting a dark object, that is back lit by thin cloud, that is in turn back lit by the sun, and moving at 100 mph plus.

Alan

Good point (it's been a while since I read their white paper on the technique). It would have to be pairs for the Bayer layout.

Being able to read the whole frame at different ISO levels would be great - but I understand the benefits come from it mitigating against readout noise. With a sensor system that is ISO invariant (Sony getting there, Canon now catching up) then the dual ISO solution shouldn't (in theory, I think) actually gain you anything. Well, something (as we're not perfectly ISO invariant), but now maybe too little to be worth it vs the resolution/aliasing downsides.

sploo wrote in post #18113541
That's what I've been posting about over the last day or so

Most of the posts (from other forums) on this subject have been from Iliah, but there is some disagreement from others. That said, given that he develops raw decoding software you'd hope he knows what he's talking about.

What hasn't been seen yet (as far as I understand) is a deliberately overexposed main image, and a proof that combining the top stop of highlights from the subframe genuinely gives you real data.

Because the subframe only contains one pixel per pair I suppose there may be artefacts (it'd be like taking a 60MP sensor, then taking highlight data from every other vertical column).

Problem is, the people who have the camera in hand first are not people who know how discern this and devise a test to get right to the heart of the matter. Most of the people that Canon let use the camera before release are people who think that what they see on the screen in Lightroom is the "RAW image". One guy was very helpful on DPreview with the 1DxII, taking suggestions from the forum, until the threads were deleted at his or Canon's request.

BigAl007 wrote in post #18113594
Actually it's line pairs, not lines, so you have two rows of pixels at one ISO, and a second pair of rows for the other ISO value. It has to be row pairs, because if you used alternating rows, you would destroy your colour information. Unfortunately because the Bayer CFA uses alternating rows of Red/Green and Green/Blue filter pairs, the full colour data is only sampled at half the sensel resolution. Because you would normally be able to effectively look at both the RGGB and the GBRG combinations you do lose colour fidelity where you have horizontally running fine detail where you have the higher ISO rows blown. It is a shame that you cannot read out the sensor multiple times, to do true DualISO recording. I would love that, even if it meant a halving of the frame rate, since I am often shooting a dark object, that is back lit by thin cloud, that is in turn back lit by the sun, and moving at 100 mph plus.

Alan

Kind of like the HDR video where it's taken at 60FPS and alternate exposures to produce a HDR 30fps video.... talk about processing to perform that task. I hadn't heard about that before but pretty cool stuff.

Global shutters to be in a new line of cin cameras in couple months. C700 cinema. http://flip.it/CsjxYk. So they coulda put new shutter in 5D and other cameras. But I guess they do not want to kill highend model sales.

John Sheehy wrote in post #18113617
Problem is, the people who have the camera in hand first are not people who know how discern this and devise a test to get right to the heart of the matter. Most of the people that Canon let use the camera before release are people who think that what they see on the screen in Lightroom is the "RAW image". One guy was very helpful on DPreview with the 1DxII, taking suggestions from the forum, until the threads were deleted at his or Canon's request.

I think I recall that thread. I can understand why manufacturers get jumpy about this stuff, but it does show the power of harnessing "the community" to really get the best out of a product.

Talley wrote in post #18113619
Kind of like the HDR video where it's taken at 60FPS and alternate exposures to produce a HDR 30fps video.... talk about processing to perform that task. I hadn't heard about that before but pretty cool stuff.

Kinda. With HDR video (like HDR stills) it's about taking two separate exposures with slightly different settings (and blending) in order to improve dynamic range. For stills the killer is movement between the exposures.

I guess you could consider the aliasing for HDR video to be temporal (i.e. you're compromising across time, due to taking multiple shots to get one image - but each image is full resolution). Dual ISO raw compromises across resolution (i.e. losing alternative line [pairs] to some degree, but getting no temporal ghosting [due to movement] as it's from a single exposure).

Man, why do you keep talking about video features, enough already

John Sheehy wrote in post #18113543
Well, Canon hasn't made a camera that has high levels of low-frequency noise and/or banding noise in a few years now. The 5D3 certainly isn't one of the best, but its non-random noise is not as intrusive as it is in some older cameras, like it's predecessor, the 5D2. I used to own the 5D2 (the only camera I've ever let go of - I still have every camera I've ever owned except the Pentax MX or ME I kept in a poorly-ventilated college darkroom, and corroded inside). I saw banding noise in SOOC JPEGs, with no pushing. Doing everything just right (IOW, wrong) like shooting at f/1.4 at ISO 250 with HTP, was a recipe for shadow disaster.

Much of the improvement that one can expect with greater base-ISO DR (high-ISO "DR" is really just a proxy for SNR for low signal) is only had by changing the way you shoot. For example, say you wanted to take a shot at night of people sitting at an outdoor cafe, poorly lit, but with neon signs in the background. You see that you need ISO 1600 to get the shutter speed you need. You can shoot at ISO 1600, and turn all the colorful neon lights white, or you can shoot with the same shutter speed and f-number at ISO 100, and maintain the color of the neon lights. 5D2 = noise disaster; 5D3=less of a disaster; 5D4 (and many current cameras) = success.

Oh, for sure, knowing I can do it now will definitely give me some more latitude to create better photos when I do my job of not blowing out highlights or underexposing shadows by 10 stops LOL But the bottom line is that I'll still be working to accurately expose like I have with every camera I've used before, and now just have a better safety net with smaller holes

It'll be fun experimenting though, that's for sure.

One of the nice surprises, at least to me, is that this camera will ship so soon after the initial announcement. I haven't looked it up to check, but it seems like this is unusually short... just two weeks.

I'm not in the market for a new camera body at the moment but I've enjoyed the discussion here. I hope Adobe and Capture One are able to release their software updates quickly so the owners of the 5D Mark IV can use post processing procedures to which they've grown accustomed. I'm looking forward to seeing the photos posted here on POTN... should be great fun.

I haven't read through the 50+ pages yet, but I think the real question is, why would you drop $3,500 when the A7Rii is $3,200? You don't need to worry about the shadow recovery being bad there...and the dynamic range is to die for. Plus it has the best AF in the world currently. I'm not criticizing you all, just asking a legitimate question as to why you aren't considering the A7Rii for cheaper.

Weather sealing ?

Seriously tho .. not getting a 5d4.. not now anyway

One of the best things about canon over sony is westher sealing and built like tanks ..long history of excellent customer service..

That being siad i just bought an a7sii. I feel the need to baby it

JM Photos wrote in post #18113657
I haven't read through the 50+ pages yet, but I think the real question is, why would you drop $3,500 when the A7Rii is $3,200? You don't need to worry about the shadow recovery being bad there...and the dynamic range is to die for. Plus it has the best AF in the world currently. I'm not criticizing you all, just asking a legitimate question as to why you aren't considering the A7Rii for cheaper.

https://photography-on-the.net …showthread.php?​p=18111920

Made it easier for you. 5D3 was perfect camera for me... then add 50 improvements and they made it the next perfect camera for me. The end.

JM Photos wrote in post #18113657
I haven't read through the 50+ pages yet, but I think the real question is, why would you drop $3,500 when the A7Rii is $3,200? You don't need to worry about the shadow recovery being bad there...and the dynamic range is to die for. Plus it has the best AF in the world currently. I'm not criticizing you all, just asking a legitimate question as to why you aren't considering the A7Rii for cheaper.

Can't take a picture with a dead battery.

JM Photos wrote in post #18113657
I haven't read through the 50+ pages yet, but I think the real question is, why would you drop $3,500 when the A7Rii is $3,200? You don't need to worry about the shadow recovery being bad there...and the dynamic range is to die for. Plus it has the best AF in the world currently. I'm not criticizing you all, just asking a legitimate question as to why you aren't considering the A7Rii for cheaper.

lol dslr thread man, dslr thread, tread carefully

technically, if you're after pure shadow recovery and landscapes, the original A7r can be had for like 1K

JM Photos wrote in post #18113657
I haven't read through the 50+ pages yet, but I think the real question is, why would you drop $3,500 when the A7Rii is $3,200? You don't need to worry about the shadow recovery being bad there...and the dynamic range is to die for. Plus it has the best AF in the world currently. I'm not criticizing you all, just asking a legitimate question as to why you aren't considering the A7Rii for cheaper.

A7Rii best AF in the world???