PDAF likely to replace DFD-CDAF on SL3

[Chaemono]

Advertisement (gone after registration)

vor einer Stunde schrieb Stuart Richardson:

The S3 most certainly does have banding in pushed shadows and high ISO. It has so much of it in fact, that it is one of the reasons I am trying to sell it. The SL2 by comparison has much less. I am not sure if it is from the phase detect sensors (I believe the S3 has phase detect, as it is an SLR, but it might be on a separate chip or done with a different technology, I do not know).

Yes, certainly. Sorry for the misunderstanding, my commentary related to base ISO and Srdjan said that the X2D doesn’t show any so far either [at base ISO].

[Advertisement]

The Leica Q3 is available for pre-order at B&H Photo Video and Adorama

[Chaemono]

vor 4 Minuten schrieb frame-it:

and it has PDAF..

 

有効画素数2420万画素フルサイズCMOSセンサー、ARコーティング
ダストリダクションはセンサーシフト式
動画は6K30p 4:2:0 10-bit、C4K 60p 4:2:2 10-bit 、フルHD 120fps 4:2:0 10-bit、フルHD60fps 4:2:2 10-bit
368万ドット有機EL EVF、倍率0.78倍、アイポイント21mm、リフレッシュレート120fps/60fps
3型184万ドット フリーアングルモニタ、タッチパネル、アスペクト比3:2

AF方式 位相差AF/コントラストAF、DFD、EV -6 - 18、AFアシストライト = AF method phase difference AF/contrast AF, DFD, EV -6 - 18, AF assist light

1728ゾーンマルチパターン測光、測光範囲EV0-18
ISO範囲100-51200（拡張で50-204800）
デュアルネイティブISO、ノーマルISO100と640、V-log640と4000、HLG400と2500
5軸手ブレ補正、5段分の効果（デュアルI.S. 2で6.5段分の効果）
シャッターは1/8000-60秒（メカ）、1/2000-60秒（電子先幕）、1/8000-60秒（電子）
シンクロは1/250秒
連写（メカシャッター）はAFS/MFで9コマ/秒、AFCで7コマ秒
連写（電子シャッター）はAFS/AFC/MFで最高30コマ/秒
バッファはRAW+JPEGで200枚以上、RAWで200枚以上、JPEGで300枚以上
USB Type-C端子（USB3.2 Gen2）
HDMI Type-A
デュアルSDカードスロット
ファン オート/オート2/高速/ノーマル/スロー/OFF
Wi-Fi、Bluetooth ｖ5.0
バッテリーライフは370枚
防塵防滴
大きさ134x102x90mm
重さ740g（バッテリー、SDメモリカード込み）、657g（ボディのみ）

 

24.2 million effective pixels full-size CMOS sensor, AR coating Sensor shift type dust reduction Video is 6K30p 4:2:0 10-bit, C4K 60p 4:2:2 10-bit, Full HD 120fps 4:2:0 10-bit, Full HD60fps 4:2:2 10-bit 3.68 million dots organic EL EVF, magnification 0.78x, eyepoint 21mm, refresh rate 120fps/60fps 3 inch 1.84 million dots free angle monitor, touch panel, aspect ratio 3:2

AF method phase difference AF/contrast AF, DFD, EV -6 - 18, AF assist light

 

EV -6 - 18, AF assist light 1728-zone multi-pattern metering, metering range EV0-18 ISO range 100-51200 (extended 50-204800) Dual native ISO, normal ISO100 and 640, V-log640 and 4000, HLG400 and 2500 5-axis image stabilization, 5 steps effect (6.5 steps effect with dual I.S. 2) Shutter 1/8000-60 seconds (mechanical), 1/2000-60 seconds (electronic first curtain), 1/8000-60 seconds (electronic) Synchro is 1/250 second Continuous shooting (mechanical shutter) is AFS/MF at 9 fps, AFC at 7 fps Continuous shooting (electronic shutter) up to 30 fps in AFS/AFC/MF Buffer: 200+ RAW+JPEG, 200+ RAW, 300+ JPEG USB Type-C terminal (USB3.2 Gen2) HDMI Type-A Dual SD card slots Fan Auto/Auto2/High Speed/Normal/Slow/OFF Wi-Fi, Bluetooth v5.0 Battery life is 370 shots Dust and splash proof Size 134x102x90mm Weight 740g (including battery and SD memory card), 657g (body only)

It’s time to see if it shows PDAF banding or striping at base ISO. Someone else has to do it. 

[sillbeers15]

1 hour ago, LocalHero1953 said:

What has the launch of a Panasonic got to do with the timeline for Leica's Q3 or SL3?

Q2’s image sensor & AF specs (hardware & firmware) identical to SL2. Both are again based on Panasonic S1R.

You think Leica is capable to develop DFD/CDAF by their own development team?

 

[sillbeers15]

It now looks like there is partial news release about Panasonic’s S5ll almost every day to get public interest from now till official product release.

On AF specs of the new S5ll, the released specs states: PDAF with CDAF + DFD.

It looks like Panasonic’s new AF will be Hybrid of PDAF & CDAF. It looks like PDAF might be used for AFC & AF for video while CDAF / DFD will remain for AFS. Thoughts?

[sometimesmaybe]

22 hours ago, sillbeers15 said:

I am not aware that PDAF require specific lens design for it to work?

when Fuji introduced PDAF to the GFX100/s, the GF110mm needed a firmware update

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

[Chaemono]

vor 24 Minuten schrieb sillbeers15:

It now looks like there is partial news release about Panasonic’s S5ll almost every day to get public interest from now till official product release.

On AF specs of the new S5ll, the released specs states: PDAF with CDAF + DFD.

It looks like Panasonic’s new AF will be Hybrid of PDAF & CDAF. It looks like PDAF might be used for AFC & AF for video while CDAF / DFD will remain for AFS. Thoughts?

My thoughts are that AF will be slower with the 60 MP SL3. You may have to wait for the SL3-S or go with the S5 II.

[LocalHero1953]

Advertisement (gone after registration)

14 minutes ago, sillbeers15 said:

Q2’s image sensor & AF specs (hardware & firmware) identical to SL2. Both are again based on Panasonic S1R.

You think Leica is capable to develop DFD/CDAF by their own development team?

 

Well, Leica certainly has had the capability, even if they didn't exploit it at the time. And the SL (2015), CL (2017) and T (2014) came out long before Panasonic's S1/S1R (2019). There is nothing in this history that implies that Leica and Panasonic are releasing identical models in lockstep.

As to where the skill base lies....Leica and Panasonic were certainly cooperating long before the L-Mount Alliance (2018), but I suspect that Leica had a significant input into the AF system development - Panasonic would have needed Leica's experience from the SL in developing such a major step forward in their product line: point-and-shoot via MFT to S1R with much bigger lenses and higher AF performance requirements.

Just as it's an interesting question where Leica would be without external electronic engineering skills, it is also an interesting question where Panasonic would be (in photography) without Leica's opto-electronic skills (designing lenses so that they work fast with AF motors, sensor etc. Panasonic has the vanilla electronic engineering, and Leica has the vanilla optical engineering, but the necessary vital skill is how you put it together - I suspect this is more a joint activity than Leica is given credit for. But we're all just guessing. 

[frame-it]

30 minutes ago, sometimesmaybe said:

when Fuji introduced PDAF to the GFX100/s, the GF110mm needed a firmware update

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

probably so that the AF motor doesn't die most lenses might probably need a FW update for PDAF

 

Edited December 29, 2022 by frame-it

[sillbeers15]

27 minutes ago, Chaemono said:

My thoughts are that AF will be slower with the 60 MP SL3. You may have to wait for the SL3-S or go with the S5 II.

There should be no excuse for a slower SL3 AF w. 60Mpx if a fast enough CPU processor plus separate AI processor like Sony is doing.

[beewee]

23 minutes ago, sillbeers15 said:

There should be no excuse for a slower SL3 AF w. 60Mpx if a fast enough CPU processor plus separate AI processor like Sony is doing.

Faster processor => More power and heat => lower battery life, image noise, video recording limits or overheating

The only way to solve the above and still get faster processing is to go to a more advanced silicon node but there comes a limit at which point either: a) you’ve reached the limits of current technology, b) you are competing with the likes of Apple and Nvidia for the most advanced nodes, or c) you lack the volumes needed for TSMC or Samsung to bother talking to you or manufacture at a reasonable price.

[LBJ2]

1 hour ago, sillbeers15 said:

It now looks like there is partial news release about Panasonic’s S5ll almost every day to get public interest from now till official product release.

On AF specs of the new S5ll, the released specs states: PDAF with CDAF + DFD.

It looks like Panasonic’s new AF will be Hybrid of PDAF & CDAF. It looks like PDAF might be used for AFC & AF for video while CDAF / DFD will remain for AFS. Thoughts?

• "Image plane phase detection AF / Contrast AF 
• DFD technology"

*  l-rumors.com

It is interesting to read these two AF specs were listed separately, so you may be on to something with your idea about applied to either AFC/video and/or AFS scenario. I can only imagine Panasonic would do their very best to make sure all their years of DFD development would not be jettisoned. Personally, I was hoping Panasonic would somehow find a way to include DFD tech with PDAF/CDAF. But also to @beewee's point, would probably require significant processor upgrade even with a 24MP-only FF camera. 

At a minimum this will be the talk of the industry. Every reviewer will be all over this Panasonic PDAF w/ CDAF and/or DFD. I certainly hope Panasonic nailed it right out of the gate (would be unusual for a first release) as the media industry interest and scrutiny will be very high and first pass results will hit the internet the minute the S5II is officially announced w/ "pre-production" caveats of course. 

[sillbeers15]

37 minutes ago, beewee said:

Faster processor => More power and heat => lower battery life, image noise, video recording limits or overheating

The only way to solve the above and still get faster processing is to go to a more advanced silicon node but there comes a limit at which point either: a) you’ve reached the limits of current technology, b) you are competing with the likes of Apple and Nvidia for the most advanced nodes, or c) you lack the volumes needed for TSMC or Samsung to bother talking to you or manufacture at a reasonable price.

Apple M2 & M1 are known to be fast but yet energy efficient comparing to Broadcom & Intel processors. Of course TSMC chips make up both M1 & M2. So faster equal more heat & energy consumption is not necessary true with 5nm in mass production & 3nm soon to be produced in new TSMC fab. Back to cameras, just look at the performance of Nikon Z9 against its specs of a high performance processor making it a industry leader in AF accuracy & speed.

[Chaemono]

Just in time for the Maestro IV processor in the SL3 😁https://www.tomshardware.com/news/tsmc-reveals-2nm-fabrication-process

[SrMi]

4 hours ago, Stuart Richardson said:

The S3 most certainly does have banding in pushed shadows and high ISO. It has so much of it in fact, that it is one of the reasons I am trying to sell it. The SL2 by comparison has much less. I am not sure if it is from the phase detect sensors (I believe the S3 has phase detect, as it is an SLR, but it might be on a separate chip or done with a different technology, I do not know).

Only some mirrorless cameras have on-sensor-phase-detection (OSPDAF). S3 has PDAF, but, as a DSLR, it does not have OSPDAF. Therefore, PDAF has no effect on banding in S3.

Most sensors show some banding in low light and when pushed very hard. Some show it earlier, some later.

[SrMi]

2 hours ago, sillbeers15 said:

It now looks like there is partial news release about Panasonic’s S5ll almost every day to get public interest from now till official product release.

On AF specs of the new S5ll, the released specs states: PDAF with CDAF + DFD.

It looks like Panasonic’s new AF will be Hybrid of PDAF & CDAF. It looks like PDAF might be used for AFC & AF for video while CDAF / DFD will remain for AFS. Thoughts?

I would assume that AF-S would also use PDAF but with a second CDAF tuning stage, like most other mirrorless cameras.

[Stuart Richardson]

In any case, I welcome PDAF in L mount if only to stop the grumbling about AF performance, haha. But I am sure many will find a way to complain about AF even if it is comparable to Sony and Canikon. BTW, I don’t discount there is a difference, not at all. It is not relevant in my work, but it is clearly critical to others. I just hope they can implement it without introducing banding or striping. AFc is not important to me, but having super malleable files is. The SL2 is great for having low or no banding or color contamination on pushed shadows, though it can get grainy. 

[BernardC]

3 hours ago, sillbeers15 said:

You think Leica is capable to develop DFD/CDAF by their own development team?

We know that Leica and Panasonic are working together for the next generation of L-Mount cameras, so the question is rhetorical. We don't know how the work is being split. The PDAF work might be done in Japan, or in Germany, or both, or somewhere else. Why speculate, it really doesn't matter where this particular chair is located. Osaka and Wetzlar are both overcast with a high of 11°C today, so not much difference!

3 hours ago, sillbeers15 said:

It looks like Panasonic’s new AF will be Hybrid of PDAF & CDAF. It looks like PDAF might be used for AFC & AF for video while CDAF / DFD will remain for AFS. Thoughts?

The main thing missing from their current system is the initial "direction of focus" information that phase-detect provides. After that DFD is just as fast, and more accurate. I think they'll find a way to use phase information without introducing too much noise. I don't think they'll rely predominantly on PDAF; they don't have to, unlike their competition. Let's not forget that PDAF is inherently less accurate, and that the difference is due to physical limitations, so it only gets worse with sharper lenses and more pixels.

[Simone_DF]

2 hours ago, LBJ2 said:

Every reviewer will be all over this Panasonic PDAF w/ CDAF and/or DFD. I certainly hope Panasonic nailed it right out of the gate (would be unusual for a first release) as the media industry interest and scrutiny will be very high and first pass results will hit the internet the minute the S5II is officially announced w/ "pre-production" caveats of course. 

It's unlikely. Adding PDAF capability is only half of the story. Software optimisation is important, and it takes time. I suspect we'll see substantial improvements with the first or second firmware update in a year or two after the initial release.

[Advertisement]

Hallo,

Look here: Leica L-Mount

[sillbeers15]

2 hours ago, SrMi said:

I would assume that AF-S would also use PDAF but with a second CDAF tuning stage, like most other mirrorless cameras.

Hummm…that was my initial thought that DFD would be redundant while the Contrast detection’s strength is accuracy as you mentioned. What puzzled me is if the information released so far is accurate, then the mention of DFD specifically would means Panasonic still utilises it for a specific role on initial focusing direction. My own personal experience tells me that DFD is not slow in driving AFC but it gets totally confused when the subject is against a back lighted background. And we all know many users complain about the pulsing effect of the AFC box in the EVF. The pulsing effect is especially detrimental in video focusing function of the camera.

So to solve the unreliable AFC problem and to resolve the pulsing effect brought forth by DFD on tracking focus continuously, likely PDAF will replace DFD in these scenarios. Just my guess.

Edited December 29, 2022 by sillbeers15

[LBJ2]

Official Lumix teaser for Jan 4th announcement: New Year New Phase!  -- l-rumors.com