Django

9 hours ago, John Matthews said:

It would seem there's an unhealthy paranoia around focus in general. Talk to a YouTuber and they'd say it's the most important point on a camera with seemingly higher standards than Hollywood... for YouTube. It's a strange world we live in.

Its quite logical really, most Youtubers shoot solo so they value things like face/eye AF. Whereas Hollywood uses manual glass and DPs have dedicated focus pullers. AF is not at all necessary.

What's "manual autofocus" by the way?

6 hours ago, ade towell said:

I think the C70 and the C200 too when shooting in RAW both have a lovely full image, the Canon mirrorless cameras like R6 and R5 unfortunately suffer from a lack of DR, they don't seem to have the full range of latitude like the C70 does  especially in the shadows. There is something strange going on in the shadows with the R5 & R6 - seeing as we've gone there this is how our friends at Cine D describe the R6...

"What can be observed is the fact that almost no noise floor is visible – it looks like lower stops are simply cut off.

Hence, it will not be possible to dig into the shadows much without losing color and creating banding – confirmed by our latitude (underexposure) test "

This is one of the frustrating things about the Canon mirrorless cameras and it is nice to get that big chunk of DR back with the S5ii. I think this tonal range helps in overall image quality.

With Canon you have to buy one of their cinema cameras to get the full range of DR, at least in video. 

Right but to also be fair to R6 that test was done before the CLog3 update that did much improve the DR.

But overall I agree with you, Canon mirrorless video doesn't have the best DR.

This is definitely a clear advantage for Sony/Panasonic and it certainly contributes to overall IQ.

1 hour ago, TomTheDP said:

The S1H lost that aggressive NR in later firmware, the GH5 had similar issue at first. The S1 never had the aggressive NR though, probably one of the better cameras in terms of an organic NR free image, it was rated at 12.2 stops. The C70 in RAW is 12.3 stops, which is very close to what the S1 is doing.

 Of course the S1 may have less dynamic range in RAW due to the likely less processing going on. But from using both cameras on projects the S1 seems to hold highlights better while the while the C70 is better in shadows. But I know people have noticed the C70 in 10 bit is doing a lot of processing, which is why I assume the dynamic range in RAW is less. 

12.3 stops in RAW but visually the C70 in RAW: "About 13 stops can be identified above the noise floor, as well as a 14th and a faint 15th."

And yeah when you switch to 10-bit CLOG2: 12.8 stops.

Both results are imo mighty impressive for an entry level cine cam. 

And with the rumoured 24 stop Canon sensor, it seems like they aren't planning to stop here!

As for the S1's NR well I'm just going by what cined said in their review:

"Looking at the noise floor and the lower steps of the step chart of the Panasonic S1, there is very little noise visible – it looks ultra clean. Although noise reduction was set to the lowest setting, we can assume that there is still a lot of internal noise reduction going on. "

"For me, if the image is too clean, it lacks mojo."

"I have to say that I wish the video images received would have a bit more “mojo” in them. The Current  image is very clean – maybe too clean, at times. "

I dunno, maybe firmware diminished the NR? Hope that's also the case with S5ii since someone was just mentioning aggressive NR (hopefully only in standard profiles not VLOG).

This is perhaps an extra reason to wait for S5IIX and have ProRes which should normally have the less NR processing.

54 minutes ago, TomTheDP said:

yeah meant to quote abe. 

In terms of the the DGO sensors it is cool to see. Apparently the URSA mini 4.6k was DGO as well. 

The only thing is the C70 and also the GH6 don't seem to have more dynamic range than say the S1's non DGO full frame sensor. I am curious to see if we'll get a full frame DGO sensor from someone. I kind of prefer single ISO sensors to dual for whatever reason. 

Actually according to CineD, S1 has about 12 stops and C70 is closer to 13. Also according to them, S1 uses aggressive NR (even when set to -5) to achieve such results. While C70 has virtually no NR with that score. Something extra to consider when comparing metrics.

Also, the nice thing about how the DGO sensor works on C70 is very clean shadow information. So low shadow noise, high DR with pleasant roll-off. That's a dream type of sensor for me personally. I'll also take that over dual gain ISO any day.

57 minutes ago, HockeyFan12 said:

But it feel like Sony and Black Magic could make a $4k cinema camera that's as good as the Alexa LF or Venice. Understandably, they probably don't really want to.

Not that easy. The Alexa ALEV sensor still holds a lot of secrets. I think Canon are the ones getting close with the DGO sensor inside C70.

59 minutes ago, HockeyFan12 said:

Why does Black Magic have great color on both the P4K and P6K? Maybe they picked the right sensors?

BM have great CS resources via DaVinci RS. But even they keep trying to improve their CS, now at Gen5.

2 hours ago, TomTheDP said:

I never liked using S35 4K mode on the S1 as it always looked softer and noisier, especially above 800 iso. It is just surprising the Canon in FF 4k mode looks worse. 

Ok, I think you meant to quote @ade towell I never spoke on that particular comparison.

4 minutes ago, TomTheDP said:

Man that speaks really badly for the Canon. 

Huh? How so? Complimenting Panny IQ doesn't mean speaking bad on Canon.

From the camera conspiracy tests, the Canon skin tones are still king and beat both A7S3 and S5ii imo. 

Panasonic may have better DR, sharpness, motion though.. I dunno, I wanna see actual comparisons with same lenses.

9 minutes ago, hyalinejim said:

Hey, don't knock my Tamron zoom! 😂

That's a lovely lens. It's like T2.9 or something which is unreal.

Another factor contributing to the organic look, in the interests of full disclosure, could well be the slightly scuffed vari-ND I've had for years. The result is a nice blooming of the highlights at open apertures, no diffusion needed!

Sorry, I know a lot of people do seem to like them! I guess that's just the lens snob in me speaking, like I said the resulted footage looks really good, so clearly the lens hits above its price point.

The old ND might have something to do with it too! I need to experiment more with filters to rub off the modern clinical photo lens look. 

3 hours ago, Emanuel said:

I don't see why you can't have the best of both worlds to couple a R5C (8K60p + 4K120fps) and a S5IIX (IBIS) as for instance, on PDAF camp, simply going along EF glass?

That is a possibility but matching footage from different brands is something I'd rather avoid, even though using EF glass on both platforms should make things bit more consistent. (however color science and overall image processing would still differ).

2 hours ago, ade towell said:

Got to say the S5ii footage in 4k apsc mode with kit lens at f5.6 1600 iso looks cleaner and to my eyes has a more pleasing image/colours than Canon R6 4k in full frame mode with L series 24-105mm at f4 1250 iso. That's surprised me - it was starting to get dark outside by the time I was using the R6 so the overall lighting had changed a bit. Tried your CST Resolve settings hyalinejim on the S5ii, and the R6 with Canon log3 to WDR lut - they both give the footage a bright glossy look with subtle differences. Sorry can't show any footage as of yet.

Yeah I mean the S5ii footage hyalinejim provided also convinces me that the 4K footage looks indeed super clean, with a very pleasing organic look. Even though it's only shot with Tamron zoom. Pretty impressive. 

I would of course be curious to see footage with same EF-L glass shot on R6 vs S5ii for a more definitive comparison.

15 hours ago, ade towell said:

Just to be clear it's the Canon R6 mark 1 I'm comparing the S5ii with, R6 mark 2 does solve some of the above issues such as 30 minute recording limit and overheating although I'm not 100% convinced by Canon on that. The Panasonic has a fan and been tested at much higher temperatures. But the R6ii is nearly £800 more expensive in the UK which puts it in a different price bracket for me

Syncro scan was also really handy as all the speakers were in front of a screen

Anyway just playing around a bit more and have to say the IBIS on the S5ii is amazing, completely different level to anything I've tried before - wow that opens up a lot of different shot possibilities and am sure will speed up b-roll process  

Apologies for the stream of consciousness...

No that's some great feedback keep it coming! I'm trying myself to decide if I'm sticking with Canon or switching to Panasonic. I think I'd have to wait for the S5iiX though (I like chunky codecs). 

My concern is still AF. From a lot of the S5ii tests I've seen, it still feels like a first gen PDAF. Initial focus acquisition has this lag and there seems to be the occasional random hunt. And if multiple faces show up it gets very confused. Now no AF is perfect, the R6's had its issues too but R6ii does now have subject only, taken straight from the cinema line (R5 doesn't even have this) which locks on a face and never back-focuses even if the face leaves the frame. That along the low RS, no crop 4K60p and overall CS makes me wanna remain in Canon eco-system. 

That said, Panasonic has its own merits so looking forward to hearing more feedback, especially compared to Canon. 

Thanks and congrats on the new purchase.

1 hour ago, TomTheDP said:

Most people like Sony these days though. But another thing is mixed lighting conditions, which is where a lot of cameras struggle. I think one thing higher end cameras are known for at least in the cinema realm is maintaining pleasing skin tones while also maintaining color accuracy. 

Film stocks definitely give a pretty stylized look though and everyone is using film conversion luts these days. 

Agreed, Sony has improved -a lot- but you're right about mixed lighting conditions, this is were Sony always fails miserably.

Technically I guess its more a white balance issue but yeah some cameras just handle mixed lighting way better than others. We all remember A7S2 zombie skin tones.

Its not a big issue with cinema cams in the higher end because the environment is so well controlled. Shooting RAW WB doesn't even matter.

If you're talking about ARRI CS, well then yeah its the best in the world. Everyone else is just trying to copy it, including myself.

I'm not a big fan of film stocks to be honest but it often helps rubbing off the digital edge of mirrorless cams so I do use film convert when I'm trying to get away from the dreaded video look. But sometimes I like the video look. Like for shooting BTS, corporate, social media or more docu/news type stuff. Social media clients hate the cinema look, too stylised, comes off as arrogant & overkill.

I did actually try CST with those settings but to be honest was not enchanted by those results. It's all very subjective of course, and I'm not necessarily looking for complete accuracy when grading, although its probably a good place to start so I'll test that further. 

I think Sony was voted most accurate color science in recent cross-platform tests. Yet in reality they are ironically often the most disliked! Reality isn't always that pretty 😉 

@hyalinejim Thanks but I'm not quite sure how to best compare/analyse those charts? From what I can tell the reds are darker and more saturated, the blues are lighter and shift towards purple, the orange looks darker and the cyan brighter, almost vivid. How all that translates in an image I don't know, CS is still kind of a mystery to me. In the sense that perfect accuracy isn't always what you want, real life can be boring, ugly even. CS is supposed to enhance that. Make skin tones look good and even etc.

Regardless, it seems pretty clear to me now form working with the files you provided and from the camera conspiracy video comparing S5II/A7SIII/R5, that the biggest issue is red channel is over saturated. 

That said I like the Panny CS generally speaking. Like I said it kind of reminds me of Leica, which is my favourite. I think the main culprit is really the outdated single conversion LUT.

14 hours ago, hyalinejim said:

Because of the rolling shutter, or because of something else?

Mainly the RS but the low bitrate and 4:2:0 might also be less robust than the 4K. Hard to tell with the wideshot train station examples but that's what I suspect. Still a nice option to have..

14 hours ago, hyalinejim said:

I assume that RS will be improved in APSC mode, so there is that option if you know there's a train due any minute!

 

That's what I was thinking too, or shoot in 4K60p to half the RS. That's what I often do on R6 who also has poor RS.

14 hours ago, hyalinejim said:

One of the questions I was interested in looking at in the test footage I uploaded was: is the 3.1x crop in HD pixel-pixel mode usable? The answer, I think is yes but only by keeping noise low by exposing at base ISO and/or overexposing to keep noise at a minimum and also stopping down to wherever the lens is sharpest.

I haven't played much with the HD pixel-pixel files as I noticed the footage looks noisy and soft.

13 hours ago, hyalinejim said:

Skintones are looking good to me using a LUT I made for the GH6:

They look good to me too but had to tweak them a bit to avoid the red channel sat from boosting red nose syndrome. 

1 hour ago, ade towell said:

This is where I think the Panasonic 1.8 primes will be great - they are all silent AF with focus breathing control and manual focus speed control.

Yeah those seem like ideal affordable compact primes, as a prime shooter its definitely what I'd go for native lens solutions.

24 minutes ago, hyalinejim said:

To clarify: In Panasonic's LUT, reds and oranges are relatively more saturated than blues and greens even though it's a fairly low saturation conversion in general. The Resolve CST is more saturated than Panasonic's LUT. But the saturation is more balanced across hues, eg; the reds are high saturation but so are the blues and greens. The gamut mapping above helps to bring some of the super-saturated colours into range. But if you prefer a lower overall look then lower the saturation.

Cool, I'm staying away from the standard Panny LUT. The Varicam ones like "Nicest" seem.. nicer.

Overall solid 4K files. Not sure about 6K open gate. Man that rolling shutter, yikes. Gotta stay away from trains!

44 minutes ago, hyalinejim said:

@Django and anyone else who's interested I've uploaded a bunch of very short clips here at different resolutions and crop factors to check out:

https://drive.google.com/drive/folders/13zSYccjrhj1WXKDNMFioxTAoQbZyuP80?usp=share_link

Many thanks! That's more than I could hope for! Already thrown a few in Resolve, looking good. Aside from the red channel but I already knew that so bringing red down a tad in hue vs sat. What lens was this shot on?

Well color science also evolves generation after generation, but the same old LUTs often remain. 

Saturated reds is something Leica is famous for so I wonder if this is their influence. It seems like their CS is closer than ever to Panasonics now with L2 image processing.

Still some issues with the red channel vs Sony/Canon.. and he goes through Gamut, CST etc:

Yeah well shooting RAW always helps when doing heavy grading. You can't compare it to 180mbps 6K HEVC.

But I hear you, there is something special, something else coming from BMD cams. The image has a thickness, an organic cinematic property, that you just don't find in mirrorless hybrids. I notice it, my clients notice it. That's why we use BMD cams on narrative, high-end advertising etc.

That said, the gap is closing. With chunky 10-bit files the grading can be pushed nicely. It's a different look, cleaner/thinner, more digital photo like. It suits plenty of applications. Not everything needs to look super cinematic. 

Lenses play a big part too. Not too many Alexa, RED, Venice footage using $500 photo lenses.

And the bottom line is specs. We're talking compact hybrids with IBIS, PDAF & EVFs. 

Getting back to Vlog, my first impressions are good. The standard "Vlog to V709" LUT inside Resolve is pretty accurate. Kind of a mild curve (I'm used to more contrast) but its a good starting point. I am surprised at what this little HEVC 4:2:0 codec can hold.

Just saw an update on Cined concerning S5IIX: 

C4K ALL-I (H), at 800Mbps and needs to be outputted to SSD.

C4K ALL-I (L), at 700Mbps and can be recorded internally.

• Recording Apple ProRes 422 HQ in HD internally

• Record 5.8K Apple ProRes externally to SSD via HDMI or USB

@HockeyFan12 Funny I was just watching his S5ii Tokyo coverage. I'll have a play at that while I await hyalinejim's files.

As for the image looking "thin" well we these are rather low bitrate 4:2:0 h265 files. Can't expect the moon.

I do expect S5iiX to have thicker image with ALL-I and especially ProRes codec. Speaking of, I'm getting conflicting info concerning S5iiX internal ProRes. Will it be at all resolutions or FHD only?

Excellent!!!

14 minutes ago, hyalinejim said:

You can assign a function button (and 14 of the buttons on the body are re-assignable) to "Image Area of Video". This brings up a small overlay with three options: FULL, APSC and Pixel-Pixel. It's quite fast to switch between them. I have this function assigned to the red record button as I use the shutter to start and stop video clips.

Ok so basically a two-step operation. What's pixel-pixel crop factor in 4K? Does that mean a 1:1 readout?

@Thpriest That's what I was thinking, the 24mm & the 50mm, with crop mode those also give you 35mm & 75mm.

Speaking of which how easy is it to switch to APS-C/S35 crop mode? Can you assign a button to punch in&out?

Also I notice no stills/video switch and only 3 custom slots. So I guess no independent custom settings?

35mm is what I've shot most photo/video with, followed by 50mm. 

But these days, if I include iPhone shooting experience, my most used focal length is 28mm.

1 hour ago, hyalinejim said:

What are you looking for in terms of specs and content? I don't have anything coming up that I can share but if I can shoot something for you easily I will.

Ace, gotta love this place!

I'd like to see how the 6K 3:2 open-gate holds up, since it's 10-bit 4:2:0. And perhaps compare it to some C4K 10-bit 4:2:2. 

As for content, I'm not going to be very fussy, whatever is convenient for you. Ideally human subject to test skin tones but any type of daylight street scene would also do fine.

Many thanks in advance 🙏

CLog was developed on C300 for indeed 12 stops and CLog2/3 arrived with subsequent C300mk2/3 offering higher sensitivity. They claim up to 16 stops in CLog2 on DGO sensors. I just think its nice to have access to these various log curves depending on what scene you're shooting. The flattest curves often raise the noise floor. Which is why Canon CLog2 is only available on certain models and often only in post when shooting RAW. 

Anyways I need to find some S5ii Vlog footage to put it through the test. Hopefully some will pop up somewhere.

FWIW I've been looking through lot's of S5ii footage on YT and found some I really enjoy CS wise. I see a certain neutrality that reminds me of Leica (no surprise considering the ties and L2 image processing). I'm also a Leica M shooter so really value that CS and I think I'm seeing some of it.

Concerning lenses, does anyone here have experience with the Panny L-mount F1.8 primes ? 

I like a good set of affordable primes and read these have a focus breathing suppression design, smooth exposure and various manual focus ring settings. sounds ideal for video use!