Ilkka Nissila

Panasonic is reading the sensor slower in both the normal and DR boost modes, explaining how they can get more DR out of it than Nikon in their implementation. It may or may not be the same sensor. In any case Nikon's compromise is different from Panasonic's and these are both legitimate choices. The ZR was under development before Nikon acquired RED and what RED know-how they added in this camera is likely in the firmware (and post-processing support in the R3D NE format pipeline). In cined's testing the latitude test shows better retention of color across exposure adjustments in post-processing when using the R3D NE than N-RAW and so it would seem that the RED acquisition already paid off for Nikon to become more competitive in the video arena, and this is not just marketing if it benefits users. What Nikon should do now is try to make the h.265 a bit more competitive so that more people who cannot handle the raw data rates can still benefit from the camera. It would be very costly for everyone to shoot everything in R3D NE to get benefits from the camera. I personally am looking forward to seeing some Prores 422 HQ material shot with the camera and see how that fares in comparison with the Z8.

Is there somewhere where we can see an example of this problem vs. another camera with a better implementation of h.265? I think it's understandable that when a highly compressed video codec is used, there is noticeable quality loss and the manufacturer is trying to mitigate this with some algorithmic processing of the data. Is it really the case that the quality of the h.265 is worse than in a previous model from Panasonic or another camera in a similar price class, or could it be a case of increasing expectations over time as we see high-quality footage using better screens more often?

How about Prores 422? Prores 422 4K at 25 fps is 433 Mbps vs. 2.3Gbps Prores RAW 6K (normal) and 3.5Gbps for Prores RAW HQ. I would think the Prores 422 on the S1II is likely to be a good intermediate sized format between RAW and h.265, at least from my Nikon experience the quality should be very good.

While I think it would make sense for hybrid cameras to offer similar "looks" across photos and video for easier presentation together, I am not really sure storing photos in log format makes sense. First, while linear encoding would waste bits due to the highlight photon shot noise making the least significant bits meaningless, this has already been corrected in compressed raw file formats such as Nikon's (technically lossy but visually lossless) compressed NEF. If I recall correctly, Nikon simply leaves out the LSBs in highlight pixels, thus saving storage space. In log video mode, cameras bias the exposure metering to produce about three stops of underexposure compared to normal SDR photos, and this leads to a lot of noise in the main subject (if there is one). It may not be such an issue for video because in video you can do temporal noise reduction which you cannot do for photos since they're individual frames with different content in each image. Usually in still photography, people want the main subject to have the highest possible image quality, and exposure metering algorithms typically emphasize the detected or selected subject and only secondarily protect highlights from blowing out. I still almost always increase midtones in post-processing by a curves adjustment, reducing highlight contrast and bringing the subject (midtones) up in brightness. For scenes that require a large dynamic range, many photographers I know of shoot a set of bracketed frames in order to ensure high SNR for each major part of the image and then merge the images with masks or other such techniques (depending on the subject). For video, exposure blending with masks is not possible but some automated DR-enhancement methods that blend two amplification levels exist in a few cameras (dual gain output). While the idea of having highlight exposure latitude is appealing, it comes at a cost in the midtone and shadow SNR and I think many still photographers would consider the outcome to be of poor quality compared to what they are used to. It's also the case that many if not most (?) still photographers use Auto ISO and manual exposure mode as their go-to exposure mode and they expect the camera in most cases to set the ISO precisely to get close to the desired brightness for the main subject as they are shooting. I often set the camera to ISO 100 or 64 and Auto ISO, lettting the camera vary ISO from 64 to 12800 to get the exposure correct and the photos near usable as they come out of the camera with minimal tweaking. This won't work for log as most of the ISO settings are unusable in log given the 3 stop underexposure built-into the approach. Yes, you can apply +2-3 stops of EV correction and then get similar results to linear modes but then the exposures on the screen will look off and it's harder to see the subject and get the correct feeling of the scene and how it would render in the photograph. I just don't see this going anywhere outside of a few filmmakers wanting look-matched still photos when video is their primary output. Still photographers outside of agency photojournalism shoot raw and that's that for the most part.

In still photography, the storage space issue for RAW is less pressing than in video and since each still image can be studied for a long time (at least in print) people can pay more attention to quality (and photographers can afford more time into editing of individual frames with masks etc. while in video it would be extremely tedious to make exposure blending or other manually drawn mask based operations on a frame by frame basis). In the early years of digital system cameras, the difference between RAW and JPEG was more obvious and people got used to RAW because the image details were better and of course the files are more editable. For video, I suspect that RAW usage will be more limited to high end where there are professional colorists etc. and occasional shooters who don't shoot a huge quantity of material. But maybe I am wrong. 😉

Different post-processing pipelines and their settings for N-RAW and R3D NE may be what is causing such differences and not necessarily different primary data in the file, unless the person making the video actually used the renaming hack. However, of course it is possible that the data are different in the files. However, sharpening images and storing them in the raw format makes no sense as the images are not in RGB format at that time. Sharpening in that phase could mess up the colors so I doubt they are doing it.

In an unregulated state, all the money will go to the owners of the AI built on stolen data (from creatives without compensation) and no working person will have money. It'll be like the 1920s again, and remember the tariffs then made the US depression spread worldwide, leading eventually to World War II. After which a period of relative decency began, until the 1980s where all the money more and more were given to the fewest of people, leading eventually to Brexit, Trump, the Russia-Ukraine war. All of these phenomena since the 1980s happened because the multimillionaires and billionaires want to have all the money and keep it too. Adapting is the same as capitulation which makes working people the equivalent of slaves. All the money will go to the techno-oligarchs and their criminal politician friends. The only way to solve the problem is to make AI models based on stolen data illegal and erase them or give due compensation to the creators of the original teaching data that was used to make the model, and tax billionaires so that they end up with only the money that a decent life requires. This would restore fairness and decency in society and good lives to ordinary people.

8K50p N-RAW Normal is 362 MB/s. Comparisons between different codecs at different frame rates doesn't make much sense since frame rate is usually specified by the application and not used just to fill a data rate quota. h.265 is available on the ZR.

So is the High ISO NR item in the Video recording menu grayed-out when selecting h.265?

It's only a 1:2 difference. You can get similar file sizes from Prores 422 HQ 4K as N-RAW Normal 6K, and the Prores looks gorgeous at least from the Z8; I would expect the same from the ZR. There could be an issue though if you want 50 or 60 fps then the quality may not be as good from the ZR as it is with the Z8 (due to lack of support of extended oversampling).

The R3D NE is only available in bitrates similar to N-RAW high quality, not normal, which is what many people seem to be complaining about.

With F-mount lenses the AF won't be great during video but it should be OK for stills for the most part. I don't think focus pulling by hand with autofocus stills lenses (F-mount) is going to be easy. Practice with the same kind of subjects helps. For video AF, native lenses do best. Most Z mount Nikkors also can be programmed for linear manual focusing. Raw video takes up a lot of storage very quickly.

The extended oversampling ON/OFF (in the Z8/9) only affects 4K h.265 and Prores 422 HQ video at 50 and 60 fps and it should not affect raw video formats.

Are they? The Z8 is 3999€ in Finland (discounted from 4799€) and the ZR is 2449€. Likely the ZR will be similarly discounted in a few years, maybe sooner. h.265 4K 25 fps 10-bit data rate is 190 Mbps on the ZR and 190 Mbps on the Z8, so the image quality should be similar. The ZR has an optical low-pass filter though and the Z8 does not, so details in the ZR footage might be ever so slightly softer. What the ZR does not have is "extended oversampling" so the 4K footage at 50 and 60 fps is likely of lower quality than the Z8 (when extended oversampling is turned on in the latter), and this indeed could be left out of the ZR to prevent overheating in the much smaller body. At 4K 24, 25, and 30 fps the h.265 image quality should be quite similar between the two cameras (apart from the OLPF difference). However, if you record 6K in h.265 this should not be an issue as there is no need to do oversampling and you can convert the footage to 4K using your method of choice in post. Summary: I can see that there would be a quality loss at 50-60 fps at 4K but the image quality should be fairly similar in h.265 between the ZR and Z8 when recording 4K 24-30 fps. This can hopefully be verified when the camera is out on the market. The Z8 is much larger and heavier which makes it less gimbal-friendly (or at least you need a much heavier gimbal than the ZR, potentially leading to body strain and discomfort, and eventually one is not able to continue). The ZR supports digital audio via the hot shoe (future products will support it, including tascam's) and 32-bit float, which are useful features for recording on-location audio. The ZR has a larger back display. I imagine only those who always record video on a tripod would be happy using the Z8 or Z9 while those who hand-hold or use a gimbal often, and those who simply need a lower-cost solution for video would prefer the ZR. The Z8 and Z9 of course are capable of high-resolution stills, with high frame rates, and have much less rolling shutter in stills shooting than the ZR (which has lower rolling shutter in video mode). So, many stills shooters will prefer the Z8/Z9 if they can afford it and justify the cost and weight.

Prores RAW has an even higher data rate than R3D or N-RAW on the ZR. Nikon lists ZR data rates for Prores 422 HQ 4K 25 fps at 820 Mb/s (102.5 MB/s) and 6K at 25 fps at 1600 Mbps (200 MB/s). Thus the Prores options for 6K are even bigger than the R3D or N-RAW, but the 4K version of Prores 422 HQ is a bit more manageable than 6K. Prores 422 HQ may be easier to edit than RAW in some ways (at least distortion and vignetting profiles are applied in-camera so there is less post work to be done). I use Prores 422 HQ 4K on the Z8 as my go-to format. I love the colors and image quality.

If I'm not mistaken, the issue is that what Nikon calls extended oversampling is not available on the ZR. Extended oversampling would mean the camera is able to resample the 6K sensor feed to 4K with full sensor width at 50-60 fps. The 24, 25, and 30 fps video is oversampled by default. So the quality from h.254, h.265 and Prores 423 HQ 4K modes should be fine as long as fps rate is 30 or below. These have much lower data rates than 6K RAW. Note that I do not have the ZR. I am just guessing that 24-30 fps 4K are oversampled without extended oversampling as it is the case in the Z8. If they are not oversampled, this could indeed be a major quality issue.

Are you sure this is safe for this camera? While it may be that just pulling the card out while the camera is on in most cases causes no harm the manufacturers do typically warn that the camera should be turned off before removing or attaching the card or there could be data corruption.

32-bit float audio in the in-camera mic as well as analog input. Digital microphone interface in the hot shoe, will be utilized by Nikon's own microphone (no cables) and TASCAM's audio adapter. Larger and brighter display. Future V-mount battery compatibility. More codecs, and presumably solved the flickering problem (since it's not mentioned in early evaluations and this camera's video first nature, I would expect they solved it). 130 grams lighter than the Z6III. LUT upload into camera for viewing. Drawbacks compared to Z6III: no EVF, no mechanical shutter, second card is micro-SD rather than SD.

Depth of field wise 35 mm full frame has more options (on the shallow side). However, the "look" has other characteristics including tonal and color quality and richness. If the MF sensor is used at base ISO and given as much light as it can hold, the SNR, tonal range, color sensitivity etc. are better than the MFT. Since most of these sensors and cameras were mainly developed for stills, these characteristics may or may not translate into video image quality. Also shooting at base ISO for stationary subjects is always possible when shooting stills (using a tripod) but because of shutter speed requirements for video, and the ability to process consecutive images and merge information from them to improve SNR, things get more complicated for video. If similar interframe/dual-gain-output strategies are used across formats, if there is enough light, and if the processing power is adequate and read time can be minimized then the MF image should be superior at base ISO. However, these things are very implementation-specific and so video image quality differences between formats do not always mirror still image quality differences.

If the Smallrig cage for the ZR is used, the bottom left side is Arca-Swiss compatible and the right side has a hole which allows the battery door to be opened and at least the CFExpress type B card should be easy to swap. This depends to some extent on what kind of construction the camera is mounted to, if there is space for opening the door, but it would be possible to do on some of my tripod/monopod heads. I have the Smallrig grip for the Zf and I haven't taken that off in ages. Card and battery access on the Zf is similar to the ZR. I think this situation is just an inevitable consequence of making the camera so small. You can plug in external power via USB-C to extend the shooting time. CFexpress cards are available at least up to 4 TB capacity so that would cover quite a lot without having to open the door frequently. I am sure Nikon will make larger ZR-series cameras that have fewer compromises in how the ports, cards, etc. are accessed.

The ZR does accept timecode input via bluetooth and can output TC via HDMI.

With dynamic range, the actual number that results from a measurement depends on a number of factors, what SNR level is used as the lower end limit of acceptable quality, and whether the SNR is calculated based on patches or pixels, or something in between. An 8K camera at pixel level will have lower DR than a 4K camera or 8K camera downsampled to 4K which in turn will have lower dynamic range than 2K native camera or a 4K/8K downsampled to 2K, and so on. Noise reduction that is applied to the footage also affects the dynamic range. Only by applying an equal test at equal final presentation resolution can a fair measurement of DR be made. In stills there are well-defined standards (PDR at photonstophotos and engineering DR by DXOMark, for example, all evaluated after resampling the images to 8 MP size) but for video there doesn't seem to be any unified standard which would result in comparable numbers.

Right, but for those of us familiar with other Nikon products and their video-related features, we can guess that the feature would be supported in the ZR. For people who are not already users of other recent-generation Nikons, it may be unclear which features are in fact supported until the manuals actually show up online. In fact some features can't be found even in the manuals.

All the Expeed 7 cameras have it so it is expected in this model as well, no need to mention a standard feature I guess.

Typical high-speed sync implementations either require the flash to fire a long pulse or a series of short pulses to match the timing of the moving gap of the focal-plane shutter. This results in flash energy loss due to most of the light being blocked by the shutter curtains at faster shutter speeds than the sync speed. When using a central shutter lens such as the Hasselblad XCD lenses, you don't lose light as you increase the shutter speed (to a point). So basically if you want to shoot at f/4 in bright sunlight, with a high-speed sync implementation, your flash needs to be several times more powerful than if using a central shutter lens. With the Hasselblad you can do this with a small flash instead of a powerful battery-powered flash which would be needed when balancing bright sunlight and flash at fast shutter speeds on a camera without either a global shutter or a central shutter. Each stop faster than sync speed loses one additional stop of light with the high-speed sync in a typical implementation. For the Fuji this would mean 1 stop loss at 1/250s, 2 stops at 1/500s, 3 stops at 1/1000s. So there is a huge difference in flash size that may be needed. Having to choose between strong rolling shutter distortion and line-skipped video (and cropping) is not really a pleasant compromise to make if your reputation is mainly based on producing images of the highest quality. I can understand that Hasselblad would simply not want to deal with the heat generated by resampling 100 MP images to (say) 8 MP (4K) since it's much more practical to make high-quality video using a camera with a smaller-format, lower-resolution sensor. The X2D II is particularly small and lightweight for a medium format camera and I can see them wanting to prioritize that over a heat-managed, larger, heavier camera that can shoot high-quality video. Hasselblad makes several reasonably affordable lenses for the X1D/X2D series of cameras. The 28/4 PP is 1899€, the 45/4 P is 1199€, and the 75/3.4 P is 2529€. Basically I would need the 28 and the 75. From what I can see there isn't much difference to Fuji lens prices. Considering what Jim Kasson revealed about Fuji GF series lenses not holding focus from shot to shot (even when set to manual focus) but instead there are slight shifts in focus that are enough to cause noticeable shot-to-shot variability in MTF in lens tests, it is difficult for me to see purchasing into their system. The slow flash sync speed just kills it for me. Another factor is that Hasselblad makes a digital back that can work with the XCD series of lenses and alternatively can be used on a view camera (with the sensor flush with the front of the back facilitating the use of movements).