Deciding closest modern camera to Digital Bolex look

[Matt Kieley]

The test videos from the still-unreleased Octopus Cinema Camera looked very natural and organic to me. Who knows if it'll ever materialize though.

 

[bjohn]

1 hour ago, Matt Kieley said:

The test videos from the still-unreleased Octopus Cinema Camera looked very natural and organic to me. Who knows if it'll ever materialize though

If I remember correctly, it's so modular that you can even choose what sensor you put in it. I've been following that project for a few years with interest but it always seems so close and yet so far. Now they say "Coming 2023" but I remember seeing Coming 2020, Coming 2021, Coming 2022 so we'll see when/if it actually happens.

[Robert Patts]

16 hours ago, TomTheDP said:

The A7S3 or FX3 also have low RS. 

The Pana S1/S1H/S5 also has pretty low RS in S35 crop mode. 

I can say that in s35 mode the S5 is garbage when recording in BRAW. Perhaps internal on the S1H is better.

[cpc]

The Kodak sensor also appears to have thicker filter dyes which results in rich color and excellent color separation. Later sensors may have been optimized for sensitivity, particularly cheaper sensors. If you look at images from the BM Pocket, they have more compressed color with hues mаshed together.

[amweber21]

On 10/26/2022 at 8:15 AM, Andrew Reid said:

Cinema DNG is the key to the Fp. Even in 8bit.

It doesn't have the squeaky clean and processed look of BRAW and ProRes RAW.

I may try the 2K on the Fp-L and see if it's any good. The file sizes in 4K are a bit of an issue.

You'll want to shoot with a 1.67x crop or higher if possible. That seems to be the point where it oversamples the image. Up from the 1.30x of UHD I'm guessing due to the extra processing required to downsample further to 1080p. I could swear the full width 1080p has been improved since I first got the camera. Maybe they snuck it in with a firmware, but the improvement is still clear at 1.67x. Higher 1080p frame rates will require it 1.67x or higher anyway. 12bit to an SD card is also nice.

[Andrew Reid]

40 minutes ago, amweber21 said:

You'll want to shoot with a 1.67x crop or higher if possible. That seems to be the point where it oversamples the image. Up from the 1.30x of UHD I'm guessing due to the extra processing required to downsample further to 1080p. I could swear the full width 1080p has been improved since I first got the camera. Maybe they snuck it in with a firmware, but the improvement is still clear at 1.67x. Higher 1080p frame rates will require it 1.67x or higher anyway. 12bit to an SD card is also nice.

The Fp-L starts oversampling around 1.37x

And looks very nice and crisp up to around 2x, maybe even a bit further

The full frame and 1.1/1.2x crops are binned but still look very nice.

I just wish it had a 2.8K RAW mode oversampled from 4K RAW to save on file size.

[amweber21]

13 minutes ago, Andrew Reid said:

The Fp-L starts oversampling around 1.37x

And looks very nice and crisp up to around 2x, maybe even a bit further

The full frame and 1.1/1.2x crops are binned but still look very nice.

I just wish it had a 2.8K RAW mode oversampled from 4K RAW to save on file size.

You can see the jump at 1.30x based on this rolling shutter graph, but you're right, it looks great all the way to 2.5x. Haven't seen any evidence for 1.67x on 1080p, outside of my own observations. Give it a try if you're doing any testing. I'd be curious if you notice anything also.

I've sent a request to Sigma for something similar. 2.5K-2.8K would allow up to 60p 12bit right around the bandwidth of current 4K 30p and a big improvement on 1080p. 2.4:1 4K should also allow full 1.0x width shooting based on the amount of data needed to be read off.

 

[amweber21]

Here was my test on 4K crops at 100%. 1.0x looks okay, but some heavy aliasing. Probably okay if finishing in 1080p. Clean at 1.3x and on. No problem in 10bit or 8bit. So I will drop to 10bit if I need to stay wider. 10bit still has fantastic flexibility.

Noticed something similar in 1080p but needed an even higher crop. Don't have any test shots to show though. The bitrate didn't seem to affect it, however. Still showed up at 10 and 8bit for whatever reason.  

[Andrew Reid]

Are those 1.3x?

You need to go 1.37x before the oversampling kicks in.

[amweber21]

32 minutes ago, Andrew Reid said:

Are those 1.3x?

You need to go 1.37x before the oversampling kicks in.

Yeah, it's with 1.30x. Anything 1.24x and lower is definitely skipped in 12bit mode.

 

Sigma's fp lead Wakamatsu gave recommendations for crops: 

For the best Resolution, x1.3 – 2.48

For phase detection AF, x1.0 and x1.53

For best dynamic range, x1.24

[Andrew Reid]

Interesting that he would give 1.24 for best dynamic range, do you have a link to the interview with Wakamatsu?

Maybe the binning / oversampling different in SSD DNG 12bit vs internal 8bit DNG?

However your 1.30x shots look pixel binned and all the same detail. Different sharpening maybe.

From memory of my tests the other month, I am 99% sure it needs to be at 1.37x and anything less is similar in detail to the 1.0x full frame.

Maybe Wakamatsu's ghost writer got the numbers slightly wrong... Someone also needs to tell him to get back to work from the motorbike track and put a LOG profile in there pronto 🙂

Wonder what the deal is with that still not being in?

And an articulated screen would also be a great edition to future entries.

[amweber21]

1 hour ago, Andrew Reid said:

Interesting that he would give 1.24 for best dynamic range, do you have a link to the interview with Wakamatsu?

Maybe the binning / oversampling different in SSD DNG 12bit vs internal 8bit DNG?

However your 1.30x shots look pixel binned and all the same detail. Different sharpening maybe.

From memory of my tests the other month, I am 99% sure it needs to be at 1.37x and anything less is similar in detail to the 1.0x full frame.

Maybe Wakamatsu's ghost writer got the numbers slightly wrong... Someone also needs to tell him to get back to work from the motorbike track and put a LOG profile in there pronto 🙂

Wonder what the deal is with that still not being in?

And an articulated screen would also be a great edition to future entries.

It was from cined's review of the fp L when it was first launched: https://www.cined.com/sigma-fp-l-review-and-sample-footage/ They were collaborating with Wakamatsu regarding the testing.

Not sure how 1.24x gives the best dynamic range, but from the testing it gives the best rolling shutter which makes sense.

In a later interview they retracted their promise on a log profile stating it was a hardware limitation, so that looks to be all but dead.

Here's a quick visual to show the crossover point on both 4K and in 1080p. 1.30x for 4K and 1.67x for 1080p. No adjustments outside of the crop function. Should show the huge fine detail increase in both.

[bjohn]

Funny we were just discussing Octopus above; they just made this announcement:

Following lots of feedback, we have decided to consolidate our camera development work so far into a new camera product better poised for ultra-low-budget filmmaking — an underserved area.

Internal RAW. 5K+ resolution. True pocket-size. Under 700€.

We will make an official announcement of the product in the new year. The current high-end 8K model is still in development for bespoke client camera requirements.

[Andrew Reid]

Looks interesting

https://www.octopuscinema.com

 

[essbe]

Maybe too obvious, but what about the EOS M with ml in 2.5k (or 2.8k) crop? The cheapest alternative atleast.

[IronFilm]

13 hours ago, bjohn said:

Funny we were just discussing Octopus above; they just made this announcement:

Following lots of feedback, we have decided to consolidate our camera development work so far into a new camera product better poised for ultra-low-budget filmmaking — an underserved area.

Internal RAW. 5K+ resolution. True pocket-size. Under 700€.

We will make an official announcement of the product in the new year. The current high-end 8K model is still in development for bespoke client camera requirements.

They also mentioned ages ago, 35 weeks ago, on instagram:

"What if we ported the OCTOPUSCAMERA system software to a Raspberry Pi Compute Module 4? Well, you might get a 12-bit 4K RAW cine camera that costs under €150 and fits in your pocket! A little companion project we're investigating while other things are brewing 😉"

https://www.instagram.com/p/CaaLmwyKNhO/ 

[Django]

Kodak CCD sensors have so much mojo vs CMOS. I'm still stunned by the IQ I get from my Leica M9 vs CMOS Leicas.

I feel Fuji carry over some of that mojo with their Xtrans film simulations but the heavy chroma NR remains an issue. Still XH2S with ProRes, Open Gate and ultra low RS (5ms in 4K60) probably gives some of the best IQ for the money.

Komodo is by far the best under $10K.

Sigma is very interesting to me sensor wise since the DP2 Merill Foveon days. Again lots of mojo not so un similar to Kodak CCD. Is there any update on their FF Foveon x3 supposedly releasing in 2022?

https://ymcinema.com/2022/08/03/sigma-ceo-the-full-frame-foveon-x3-sensor-will-be-ready-this-year/

[Grimor]

My vote is for Cinemartin 🤪

 

[amweber21]

8 hours ago, Django said:

Kodak CCD sensors have so much mojo vs CMOS. I'm still stunned by the IQ I get from my Leica M9 vs CMOS Leicas.

I feel Fuji carry over some of that mojo with their Xtrans film simulations but the heavy chroma NR remains an issue. Still XH2S with ProRes, Open Gate and ultra low RS (5ms in 4K60) probably gives some of the best IQ for the money.

Komodo is by far the best under $10K.

Sigma is very interesting to me sensor wise since the DP2 Merill Foveon days. Again lots of mojo not so un similar to Kodak CCD. Is there any update on their FF Foveon x3 supposedly releasing in 2022?

https://ymcinema.com/2022/08/03/sigma-ceo-the-full-frame-foveon-x3-sensor-will-be-ready-this-year/

The latest I've heard if from an interview a couple weeks ago. Around the 9 minute mark: https://www.youtube.com/watch?v=mY672OZBrp0

He says they are in the middle of the second stage, and hope to begin the third stage by the end of the year or early next year. Camera body design won't happen until after the third stage. I wouldn't expect it until late next year at the earliest. I just hope it allows for some sort of video capture, especially if it can be CDNG  like their fp system. foveon+cdng would be the closest thing to film in my opinion.

[tupp]

On 10/25/2022 at 2:28 PM, tupp said:

Global shutter and S16 optics are significant to the look of the D16.

On 10/25/2022 at 5:47 PM, BenEricson said:

It really depends what people define the "film look" to be.

I was referring to the look of the D16 -- I wasn't referring to film emulation.

 

 

On 10/25/2022 at 5:47 PM, BenEricson said:

... it has the caked on color look due to the CCD sensor. There's also nice look noise and really nice baked in looks.

When's dessert?

Actually, "caked-on" color is a good way to describe the look from some CCD's (but not from all CCD's -- consider look of the striped CCD of the Pany-Genesis/Sony F35).

 

 

1 hour ago, Grimor said:

My vote is for Cinemartin 🤪

That's a real funny joke.

However, Cinemartin rapidly achieved working prototypes of a cinema camera.  They preceded the Octopus camera (still in development, but which is discussed with interest in this very thread) by demonstrating the use of a similar (but earlier) Ximea sensor module in a cinema camera.

What camera developments have you achieved?