Kino

On 11/29/2024 at 5:36 AM, Davide DB said:

 

Not to nitpick but.
The photos are wonderful and so are the almost static shots but the cheetah chase doesn't seem like a good choice to advertise the autofocus. There are several moments that are heavily out of focus.

I have seen much better done by wildlife operators with Red and manual focus.  It seems to me that the operator's handle is beyond question.

Edit: I'll put my hands out: they might be in focus but they look blurry because of the heat emanating from the ground. IMHO not a good example anyway.

I agree and I don't think it is the heat from the ground. There are definitely a few out-of-focus moments in the cheetah chase. Perhaps it is operator error, wrong settings, or preproduction issues. I guess my point is that, especially when considering the autosport clip, the A1II's AF is much improved from the A1, where video AF was more limited in terms of AF types, performance, and subjects.

For wildlife video, the R1 seems like the best AF in the business:

Using a DNG converter, you could also use the R1's 14-bit RAW photo bursts as video clips, since the high-speed buffer is sustained for a good length of time (10-20 seconds).

The greatest thing about the A1II is that Sony didn't replace an amazing sensor for video with a higher megapixel sensor that would create rolling shutter problems and downgraded video specs. That was my biggest concern.

You can also see the advantage of the A1II's new AF for wildlife, sports, and motorsport video:

52 minutes ago, Kino said:

Using the Gerald Undone "hack" (internal recording set to 8K, but HDMI output set to 4K), you can send out an oversampled 8K image to the Atomos and record it in 4K ProRes 422 10 bit. This bypasses noise reduction and LongGOP compression in the internal files, but I think this looks the worst in terms of color depth, as Sony lists 8K HDMI output as limited to 8-bit in the spec sheet:

Oh wait! I realized that the 8K internal / 4K HDMI setting can be recorded in true ProRes 422 10 bit, as it is bypassing the 8-bit limitation on the 8K HDMI output. So, it is truly a "hack."

On 11/23/2024 at 6:40 PM, Ninpo33 said:

I fell in love with the footage shown at the link below. There’s a trick people discovered about sending out 8k to a Ninja recorder but recording in 4.2k Gerald Undone discusses it at length on one of his A1 update videos. it gives you the best possible image out of the camera I guess. 
 

 

Using the Gerald Undone "hack" (internal recording set to 8K, but HDMI output set to 4K), you can send out an oversampled 8K image to the Atomos and record it in 4K ProRes 422 10 bit. This bypasses noise reduction and LongGOP compression in the internal files, but I think this looks the worst in terms of color depth, as Sony lists 8K HDMI output as limited to 8-bit in the spec sheet:

As for the ProRes RAW in the A1, it is hard to pick out which shots are internal and which are ProRes RAW in this video (she lists the RAW shots in one of her responses):

I always look for Blue-channel clipping as a giveaway for ProRes RAW, but otherwise everything cuts together seamlessly. This suggests that the 12-bit 4.3K ProRes RAW upscales nicely to 8K and that the 8K's 10-bit internal is good enough to match with 12-bit color.

17 hours ago, Ninpo33 said:

That reminds me, does anyone know if Sony and Atomos were ever able to get the 8k output working over HDMI? It took  2 years of them working with Canon to get it working. Haven’t seen anyone mention if the mk1 has been updated or if the mk2 has the option for 8k external. 

According to Sony's website, the HDMI output on the A1II is the exact same as the A1:

7680 x 4320 (29.97p / 25p / 23.98p), YCbCr 4:2:0 8bit / RGB 8Bit

4332x2446 (59.94p / 50p / 29.97p / 25p / 23.98p), Raw 16bit

Also, note that the gamma was locked to Slog-3 during RAW output. I believe it is still the same.

The price of entry may be higher than Canon and Nikon mirrorless, but the image quality is undeniable. That is a world-class sensor that shares tech with the 8K Venice and Burano. Coming from Canon, there is also the issue of third-party lenses, where Sony has a huge advantage that balances out the higher initial price point.

In terms of the "subsampling," this is 4K/50 from the A1 and it looks incredibly detailed for what it is:

Here is the A1 in 4K/120 (HVEC 200Mbps 4:2:2 10 Bit):

I'm just amazed at what Sony achieves in this little camera and its highly efficient/compressed codecs.

21 minutes ago, IronFilm said:

ah my response you were replying to: 

 Was me making a perhaps too subtle joke, as I was writing AImk2 instead of A1mk2

Because AImk2 is going to totally change the game for the stock footage business! (of insects, and everything else)

I saw that. I thought you were referring to the extra processing chip that allows for the new AI and AF in the A1mk2. That damn AI stock footage will finish us all!

On 11/21/2024 at 3:00 PM, eatstoomuchjam said:

Sony seems to have mostly skipped 6K for their cameras (unless there is one I'm forgetting).  They have 4K and 8K available.

I guess you are referring to Sony's consumer and prosumer cameras. This one only costs $60K!

On 11/20/2024 at 9:39 PM, IronFilm said:

True. But not because of the A1mk2, but because of AImk2

With all the extra processing and heat generation, the body should have been enlarged to match those of other flagships (e.g., Canon R1, Nikon Z9). In room temp, the A1 could record 8K up to 90 minutes without any problems, but the best I've seen for the A1 II in similar conditions is 60 minutes.

The main improvements for video over the A1 include the improved IBIS (8.5 vs 5.5), stabilization modes, rear screen (resolution and functionality), AF performance, LUT support (importing function), high ISO performance, higher second-base ISO in S-Log3 (5000 vs 4000), and the new implementation of S-Log3 to match the Venice/Burano color space and detail (as a dedicated B-cam).

Also, the 8K insect 🦋 stock footage business will never be the same!

8 hours ago, PannySVHS said:

Hard call for me to get exited about the footage above. Color grading seems the devil here. If the C80 looks as good as the C500II, it is a compelling image taking powerhouse. FX6 would be a direct contender, which works pretty well as a shoulder cam. Looking forward to rigging tutorials for the C80.

This French technical review with available English CC compares the C400 with the C500 II, FX6, and RED KX.

Considering the new triple-ISO, the C400 has better ISO performance than the C500 II. DR (14 stops usable in RAW 6K) and color fidelity ("Delta E" at two different exposure levels: lower score is better) are very similar between the two cameras.

Unfortunately, there is no real-world comparison here, but at least you can get an idea of how the C400/C80 sensor should perform in challenging conditions.

On 9/26/2024 at 7:51 PM, IronFilm said:

Nikon needs a camera that sits in the middle between stills hybrid cameras and cinema cameras. 

Whatever is the equivalent of Nikon's FX30/FX3/C70/R5C could be easily released under either the Nikon or the RED brand name. (or be released under both at once?)

Aside from a few action cams and a camcorder, let's remember Nikon's last dedicated video camera was released in 1982:

Whatever they release that is not a mirrorless/hybrid will likely be under RED. If it does have 16-bit RAW (REDCODE), it will need to deal with the heat management and power requirements. Perhaps they could do this in a Z9-size body with the addition of active cooling. There is nothing smaller that could handle 16-bit RAW with current sensors, ASICs, batteries, and the rest.

I never stated that the R1 would be a better on-set cinema cam than the C80. I actually said the opposite way up there. I just think it could be an alternative for those also interested in the C80 (like me) who need something mobile.

Then we had the discussion about why you would get an R1 over an R5 II, which I did not initiate. I only responded to the R5 fans (pun intended) with all those graphs and charts to suggest that the R5 II does not really improve on the highlight clipping problems I have with the R5C. The point here is that the 6K sensor in the C80/C400 is going to offer a substantial IQ improvement on all the R5 cameras, which really can't be compared with proper Cinema EOS.

In the meantime, I've recently seen more RAW C80 footage and it is pretty awesome:

The C80 also looks fantastic in XFAVC 10 bit. Here are some XFAVC files available for download:

Despite the charts I posted above, in most scenarios the differences between the C80 and C400 will be minor. Perhaps you will see some differences in extreme low-light with noise in RAW LQ, but otherwise the C80 looks gorgeous overall. I don't use HFR much, so that C400 advantage doesn't worry me. In any case, RED and Sony should be scared.

42 minutes ago, eatstoomuchjam said:

Great.  So since we agree that most of the discussion is useless until the cameras are released and in the hands of real users, you can stop the lengthy rants and pasting screenshots from YouTube and reviewer websites and demanding that one camera (that you have never actually touched) is better than another camera (that you have never actually touched).

Again, you like to put words in my mouth. I made no such "demands." I posted test results on the R5C, a camera I own and use. When it comes to C400, C80, and R1 comparisons, everything on this thread is speculation. Some are very offended that the R1 should be considered here or that it may turn out to be a better video hybrid than the the R5 cameras. That's not my problem.

2 hours ago, eatstoomuchjam said:

Otherwise, beyond some chat about specs, it's really worth waiting for the camera to get into the hands of real users.  Then the truth will be understood.

As for me, I'm going to guess that we'll see a lot more owner-operators and documentary crews running around with C80's and C400's than we will see them using R1's.  The R1 will likely be a niche camera, used primarily by sports photographers.

I never stated otherwise. We need the R1 and its CRM files to make real assessments about how it performs in relation to the C80/C400. It could be that the faster sensor readout on the R1 undermines any DR advantages seen in all the C400 footage and tests. The AF circuitry is also different. But I can't imagine that Canon spent the last 2-3 years designing completely different and new 6K BSI sensors to be released at the same time in such low-volume cameras.

Also, no one in his right mind who needs a proper cinema cam should go for the R1 over the C400, even if they share a similar sensor and image quality. It just becomes interesting when comparing the R1 with the C80, which is still using SD cards, and lacks the FF 6K 60p RAW, IBIS, and EVF. For those who need portability in the field, which is what the C80 is supposed to offer, the R1 is just another choice and obviously a better hybrid.

7 hours ago, gt3rs said:

Number game interpretations. For RAW with no processing you should compare to the max DR possible so it would be 14.9 vs 13.4.

The proof that in 4k Log R5II 13.6 stops of usable DR vs 12, indeed disappointing 😉

And view that you are disappointed can you point hybrid cameras with more than 13.6 stops? I believe only A1 has 0.2 more...

But again, I don't care about the numbers too much as I just open the files from both camera that I own and is definitely visible the better DR.

On Imatest results, the acceptable signal-to-noise ratio at .5 "Medium" will provide the usable DR. Otherwise, using your methodology of max DR, the R5C 8K RAW (16.1 stops) comes out way ahead of the R5 II 8K RAW (15.1 stops) when both are placed on a 4K timeline:

There is no way you can recover the max DR (16.1 stops) on an R5C. So max DR is often pointless, except for camera marketing purposes. I would rate the R5C at about 11 stops, which is confirmed by the usable DR above (10.8). That would mean the R5 II (11.7) is giving us almost 1 stop more than the R5C when 8K RAW is placed on 4K timeline. However, by the same methodology, the R5 II has no advantage (both were about 11 stops) when looking at the 8K RAW source on an 8K timeline.

I absolutely agree with the eye test and everyone should just go with that. As I stated above, it is obvious the R5 II has improved noise performance in the shadows, except at high ISOs where the R5C seems better.

What I'm not seeing is any improvement in the highlight roll-off, which is a problem with the R5 cameras. The Sony A1 by comparison has better highlight roll-off than any R5 camera, including the R5 II.

The new 6K sensor in the C80/C400 (and possibly in the R1) should demonstrate significant improvement in all these areas.

7 hours ago, herein2020 said:

Pixel peepers are going to pixel peep...and use their pixel peeping to justify their desires and purchases. 

@Kino just get whatever works for you, none of your concerns are my concerns, the R5II is the better fit for me in every way vs the R1. Also, I am as far from a pixel peeper as it gets and even I can tell in the R1 vs R5II screenshots that the R5II is at least 0.5 stops over vs the R1.

And who cares about a magenta shift? All of my footage has a green shift from my vND filter, takes two seconds in post to shift it back, or you can even shift it back in camera with a custom WB tint setting; who knows, the "magenta" shift might even counter my vND's green shift which means it all works out.

I have no intention of buying an R1. It's just an idea for the future. For VND, you shouldn't be getting any green shift. On the R5C, I use the Freewell V2 Hybrid VND/CPL (3-7 stop) and it is really impressive in preventing color shift. It is magnetic, easy to use, and comes with a nice case. You should try it.

18 hours ago, eatstoomuchjam said:

It sounds like you really like the R1, even to the point of obsessing about micro-differences between images which would be removed during even basic color grading.  Seems like you should get one.

Not really. It is difficult to recover gradations of tone and color that are not in the original RAW file. As I mentioned, I will need the CRM files to confirm. The R5 II and R1 have very different sensors that should produce visible differences in DR, RS, noise/texture, and color. Since the R1 sensor is likely a variant of the 6K sensor in the C400/C80, I would expect better performance for video than the R5 II's 10.9 stops of usable DR and 17ms rolling shutter in 8K mode (as demonstrated in the tests above).

In the following side-by-side comparison between the R1 and R5 II in RAW video, you can see the blown highlights in the R5 II when looking at the white beam above. That may be caused by a higher base ISO, but the overall DR does seem reduced here. In addition, the R1 retains rich and more accurate colors in RAW. It has better color and tonal separation, whereas the R5 II looks bland and overly magenta. These are just preliminary observations. I'm sure there will be proper tests when the R1 is released.

14 hours ago, gt3rs said:

Canon claims 16 stops for both so not sure why you are expecting significant improvements from the r1 vs r5 II. 

In my experience shooting R5c and R5 II (I own both) side by side the R5 II has better DR. The numbers also confirm this. 

On the IBIS issue I agree with you. Same for the overheating.

Btw I owned 1D III, 1Dx, 1Dx II and 1Dx III and I'm so happy that R3 and R1 does not look any better than the R5 II so I can skip those huge body that are a pita on gimbal and heavy. I use big lenses like the 100-300 2.8 on the R5 II and compared to my sports steup of years ago 1Dx III + 200-400 is much lighter and compact for a better picture quality.

I am expecting a DR advantage for the R1 (owing to the larger pixel pitch and better signal-to-noise) as well as a significant improvement on rolling shutter over the R5 II's 17 ms in 8K mode, but we don't have the tests to prove it just yet. The R1 has a near global shutter sensor readout speed, so that may negatively affect DR. Based on what I have seen of the two cameras so far, I really prefer the look of the R1 sensor, which is likely shared in some way with the C400/C80. I'm also not a fan of the magenta bias I've seen in the R5 II footage, but that is just a personal preference.

For the R5 II vs R5C, I mentioned that the former definitely has an advantage in the shadows:

In my experience, NR works well on the R5C 8K when required, so the shadow noise has a solution. Unfortunately, the R5 II does not solve the problem with highlight roll-off. For everyday use, these cameras are nearly identical:

G. Undone's Imatest numbers for 8K are also very close for usable DR (Medium).

11 stops on the R5C interpreted in CLOG3:

10.9 stops on the R5 II interpreted in CLOG2:

R5 II offers very little DR improvement on my R5C (and only in the shadows), whereas the the R1 looks like a significant improvement in both highlights and shadows. However, I will need to see more footage when the camera is released. FF 6K RAW is really the sweet spot for video. The EVF and AF on the R1 are also world class and there is no need to attach a fan/battery compartment to avoid overheating in our L.A. summers. Canon 1-series cameras are simply built to different standards in terms of weather sealing and durability.

Unfortunately, there is no anamorphic de-squeeze on the R1 or R5 II, which is one of my favorite features on the R5C. With either of these cameras, I would need an external monitor. IBIS wobble is another problem that the R5C avoids, since it has no IBIS. Perhaps, there will be an R5C II, but I'm not hopeful.

58 minutes ago, eatstoomuchjam said:

In my experience, Canon RAW ST is not "much higher quality" than Canon RAW LT.  I'd call it "nearly visually indistinguishable from Canon RAW LT."  I'm told that it can matter for special effects, but for almost anything else, RAW LT is generally enough.

As to the rest, the user interface on my EOS R5 is pretty decent for a hybrid camera, but if I'm going on set, I'll take my C70 over it nearly every single time.

i'm sure the R1 will be fine and people will make great stuff with it - but speaking for myself, I'll be a lot more excited for the C80 to hit the used market.

So far on my R5C, I've shot everything 8K RAW LT and it looks great. Since it is 8K, the data rate is just over 1000 Mbps. I'm just concerned that the 640 Mbps data rate is very low on the C80. You can see the loss of detail in the CVP test of the different RAW variants in the C400 test. The drop off between ST and LQ is significant:

6K RAW LQ is the best available quality on the C80. While it is the obvious choice for taking on set (over a hybrid), it is a disappointment that they used SD cards and limited the 6K RAW to LQ and 30p. If on-set use is part of your requirement, I just think you are way better off with the C400. It has many advantages over the C80 at a $2500 premium. Then again, I still own the C500 and barely ever use it these days . . .

It will be interesting to see how the R1 compares with the C80 and C400. According to Canon, it is also rated to 16 stops of DR in video:

https://www.newsshooter.com/2024/07/17/canon-flagship-eos-r1-announced/

The use of CFexpress Type B allows for much higher quality 6K RAW than the RAW LT found on the C80, which is also limited to 30p unlike the R1's 60p. The R1's Standard RAW is basically on par with the RAW HQ on the C400.

The R1 will retail for $800 more than the C80, but that might be worth it. You lose the I/O, NDs, and cinema features (e.g., waveforms, anamorphic de-squeeze), but you gain better RAW quality, IBIS, AF, EVF, and hybrid usability.

11 hours ago, MrSMW said:

To qualify, I mean an actual proper collab camera, ie, RED lut yes, ✅ Z Mount for Komodo camera, maybe… but an actual RED/Nikon cinema camera which is what some hope for? Dunno about that…

You will have to wait for DSMC4, which could happen in 2025-26. Aside from high-end cinema cameras, I think the next logical step for RED after DSMC3 is to push into the mirrorless market, which is where all the action is. I would expect more RED features and branding on Nikon mirrorless cameras.

I was thinking of Scott Balkum, not Tim Daust: