Rishi Sanyal

Rishi Sanyal

DPReview Administrator
Lives in United States Seattle, WA, United States
Works as a Deputy / Science Editor
Has a website at www.rishi.photography
Joined on Feb 25, 2014
About me:

Although I'm a scientist by training, having recently completed my Ph.D in biophysics, photography has always been a huge passion of mine. When I saw an opportunity to meld the two at DPReview, I jumped at it... and now here I am!

Comments

Total: 1346, showing: 1 – 20
« First‹ Previous12345Next ›Last »
On Sony Alpha 7S Review preview (467 comments in total)
In reply to:

Denialisnotariver: Personally I would rather see how things look pushing 1, 2, or 3 stops using various ISO settings and not some target of 6400. I recall my 6D seems to have a stop and a half ability at ISO 800 that still looked reasonable in the shadows. Seeing one stop push from 200 Vs. 400 and the like would be more intersting to me. I fail to see how looking at the comparison of ISO 6400 as the target ISO is very useful given that noise is always more likely there in low light. To me varying the base ISO and then pushing that a few stops would seem more telling,but then the testing illumination would of course have to change. Using low ISO settings and decreasing light and increased exposure times would be intersting too.

There's still some noise even in the shadows there at ISO 800 after the push (relative to ISO 6400), so you *are* still paying a bit of a noise cost, and that's exactly what that test is designed to show.

It's just that it's a small noise cost, at that point, b/c so many tones have been pushed above the downstream read noise floor by ISO amplification, that the downstream read noise has less and less impact.

That's why you find the pushes reasonable at higher ISOs on the 6D. The point of our ISO-invariance test, though, is to show you even the faintest bit of noise cost you might be paying by pushing ISO X by Y EV instead of just dialing in ISO (X + Y EV) in-camera.

Direct link | Posted on Mar 30, 2015 at 04:28 UTC
On Sony Alpha 7S Review preview (467 comments in total)
In reply to:

Denialisnotariver: Personally I would rather see how things look pushing 1, 2, or 3 stops using various ISO settings and not some target of 6400. I recall my 6D seems to have a stop and a half ability at ISO 800 that still looked reasonable in the shadows. Seeing one stop push from 200 Vs. 400 and the like would be more intersting to me. I fail to see how looking at the comparison of ISO 6400 as the target ISO is very useful given that noise is always more likely there in low light. To me varying the base ISO and then pushing that a few stops would seem more telling,but then the testing illumination would of course have to change. Using low ISO settings and decreasing light and increased exposure times would be intersting too.

Btw our new test showing the result of other base ISO pushes are in the E-M5 II review, & look for it in future reviews (a7 II, for example).

With the 6D, btw, a stop and a half EV push sounds perfectly reasonable, but again - it's which tone you're pushing that matters. By ISO 800, actually, you won't have many tones near/below the noise floor b/c every tone has been amplified 8x relative to base ISO. Hence, only the really, really dimmest of tones will still be below the (relatively high) downstream read noise floor of the 6D, and so you'll generally find pushes to be more reasonable. You can actually analyze this yourself [here](http://www.dpreview.com/reviews/image-comparison/fullscreen?attr134_0=canon_eos6d&attr134_1=canon_eos6d&attr134_2=canon_eos6d&attr134_3=canon_eos6d&attr134_4=canon_eos6d&attr134_5=canon_eos6d&attr136_0=1&attr136_1=2&attr136_2=3&attr136_3=4&attr136_4=5&attr136_5=7&normalization=full&widget=197&x=-0.3147520322641629&y=0.38478566738579867).

Direct link | Posted on Mar 30, 2015 at 04:26 UTC
On Sony Alpha 7S Review preview (467 comments in total)
In reply to:

Denialisnotariver: Personally I would rather see how things look pushing 1, 2, or 3 stops using various ISO settings and not some target of 6400. I recall my 6D seems to have a stop and a half ability at ISO 800 that still looked reasonable in the shadows. Seeing one stop push from 200 Vs. 400 and the like would be more intersting to me. I fail to see how looking at the comparison of ISO 6400 as the target ISO is very useful given that noise is always more likely there in low light. To me varying the base ISO and then pushing that a few stops would seem more telling,but then the testing illumination would of course have to change. Using low ISO settings and decreasing light and increased exposure times would be intersting too.

Thanks for your feedback. We are now starting to push base ISO shots 1, 2, 3 EV, etc. to answer, hopefully, the question you're after. Even then, though, our tests still won't be *comprehensive*, since it's not the # of stops of pushing that is the relevant question here, but rather *which tones* are able to be pushed. This is a subtle point, but an important one. When a camera has a high read noise floor (e.g. 5D Mark III), tones that fall anywhere near it are the ones that can't be pushed. Tones well above this floor, though, can be pushed 3, 4, 5, 6 or however many stops without showing much difference compared to a camera with a low (downstream) read noise floor.

To get at the other question you're asking - pushes of other ISOs, our ISO-invariance test is actually a good proxy as it shows you the ISO beyond which there's no more benefit to performing the gain by increasing ISO in-camera. Doing it in post, of course, allows you to preserve significantly more highlight tone.

Direct link | Posted on Mar 30, 2015 at 04:13 UTC
In reply to:

Lassoni: I really don't get it why this stuff isn't implemented in-camera yet. It shouldn't be hard when the camera has a contrast detection mode to compare the pictures / data with to phase detection.

I don't disagree. AF systems in DSLRs need some serious disruption in my opinion, or I wouldn't be surprised if some on-sensor PDAF implementation eventually supersedes traditional PDAF. Inaccuracy and imprecision of traditional PDAF is starting to look pretty unacceptable next to what some mirrorless cameras offer (though of course the latter lags in other ways).

Direct link | Posted on Mar 29, 2015 at 00:39 UTC
In reply to:

AlexisH: One more thing you don't need with a mirrorless camera. Although I guess they could still finely tune auto focus to account for focus shifting.

And my point is that it's just as much a crutch for Canon or Nikon as it is for Sigma. I'm having trouble understanding why you don't get this. There are many, many on-brand lenses that require adjustment, and which require different adjustments with different copies of bodies.

It's a reality of how the PDAF systems work - a reality you're denying by somehow implying that Canon and Nikon on-brand stuff are immune to this issue.

Direct link | Posted on Mar 27, 2015 at 00:06 UTC
In reply to:

AlexisH: One more thing you don't need with a mirrorless camera. Although I guess they could still finely tune auto focus to account for focus shifting.

I think what you're leaving out of this entire discussion is that it's *not* just a patent thing, since Canon lenses themselves need calibration on Canon bodies, and Nikon on Nikon bodies, etc.

That rules out the patents as the only, or even major, issue here.

In fact, there are many other things at play. Manufacturing tolerances of mounts, lens elements, etc., for example, that just cannot be accounted for, unless you send *your* body and lens in for adjustment. That's just a reality of the way the tech works. Furthermore, residual spherical aberration in lenses causes more corrections to be made between PDAF sensor measurements & image forming light, & this may not always be correct. It's complicated.

And I think you entirely discounted our evidence that on-brand lenses also tend to need calibration. And I haven't seen any collated statistics yet that show Sigma lenses have more issues than on-brand lenses, as I find reports of both Sigma and on-brand lenses having issues.

Direct link | Posted on Mar 27, 2015 at 00:05 UTC
In reply to:

raincoat: How does this compare with
http://michaeltapesdesign.com/focustune.html

I have used neither, but have a LensAlign target widget

Not quite as automated, but LensAlign (which we also use from time to time) also works very well. But it doesn't control the camera. What'd be interesting to see would be if the two different methods agreed with one another.

Direct link | Posted on Mar 26, 2015 at 23:58 UTC
In reply to:

Jan Madsen: I purchased the previous version. A good idea, but unfortunately a completely useless system. Far too many bugs, and when I eventually got a measurement / AF correction it was often way off for large aperture lenses - where it really counts. For F/2 and smaller apertures it works ok, for f/1.2 the results fluctuates, and for the Canon 50mm f/1.0 it is almost impossible to get anything meaningful out of the system.

Now if only Reikan would offer support, and correct the numerous bugs, fine, but this company has offered the worst support I have ever encountered with a commercial product - that is none. You can create a support case, and then - nothing. No answer for months. Only way to get any kind of reply from them is to plaster their Facebook wall with notes about the poor service - that worked twice, but then silence again. Unless Reikan has made a dramatic change of attitude towards the customers I would say use other options. Sadly.

Yeah the AF consistency stuff is very exciting. There's just so much data there to be mined.

Thanks for sharing your experience with the 85/1.2. Did you test consistency with it? I'm very curious about consistency of that vs. the Canon 85/1.8.

Direct link | Posted on Mar 26, 2015 at 23:50 UTC
In reply to:

Mais78: I have focal and it is useless, totally inconsistent. The is MUCH MUCH more accurate and effective for Canon bodies (and free...)

http://arihazeghiphotography.com/MA-web/

Not quite as automated, but that's interesting. Are those focus shift units equivalent to AFMA units? I'd be very curious.

Direct link | Posted on Mar 26, 2015 at 23:48 UTC
In reply to:

AlexisH: One more thing you don't need with a mirrorless camera. Although I guess they could still finely tune auto focus to account for focus shifting.

Incorrect. The dock fixes back/front-focus. So does AF Fine Tune (Nikon) and/or microadjustment (Canon).

Why are you incorrectly claiming it's a non-solvable issue? Sigma's dock even allows you to calibrate/micro-adjust *better* than any Canon or Nikon lens. That's literally an *advantage* Sigma Art primes have over native primes.

We get hundreds of lenses/bodies through here at DPR. We try & calibrate every single one if it's got shallow DOF. Just the other day, a Sony lens on a Sony body needed +19 AFMA, a Nikon on a Nikon needed +10, a Canon on a Canon needed -8, etc.

In fact, probably more combinations *need* microadjustment than don't.

So please stop spreading false claims that calibration is somehow more of a crutch for one brand than another. I *might* believe that Sigma lenses have more back/front-issues prior to calibration than native lenses- b/c they don't have access to all the same data the native brand lenses do. But I'll believe it more after a stat sig analysis.

Direct link | Posted on Mar 26, 2015 at 23:02 UTC
In reply to:

Ross the Fidller: "Olympus's forthcoming 300mm F4 lens offers a focal length equivalent to 600mm. Although not as fast as a 600mm F4 would be for full-frame, it's highly portable by comparison."

Saying "Although not as fast as a 600mm F4 would be for full-frame" in the above is just nonsense!

To Jylppy's point: lenses for smaller sensors with equivalent resolution need higher resolving powers than similar lenses for a larger sensor. That's just plain fact.

Direct link | Posted on Mar 26, 2015 at 21:50 UTC
In reply to:

Ross the Fidller: "Olympus's forthcoming 300mm F4 lens offers a focal length equivalent to 600mm. Although not as fast as a 600mm F4 would be for full-frame, it's highly portable by comparison."

Saying "Although not as fast as a 600mm F4 would be for full-frame" in the above is just nonsense!

@badi: if lens speed is refers to the 'maximum aperture diameter', then what we've said is absolutely correct, as a 600mm F4 lens has a 150mm diameter while a 300mm F4 lens has a 75mm diameter.

I don't think any of us are really disagreeing with one another here.

@cinemascope: "Sure I can use a 2.8 lens wide open on FF and be happy with the DOF, but then we are gathering less light, etc..."

Confused - what's gathering less light?

Also, folks should keep in mind that FF lenses are being updated, whereas mFT lenses tend to be newer (as it's a new system). New FF lenses tend to take advantage of the higher pixel densities of newer FF sensors, while still not needing the resolving power demands of smaller sensors to reach equivalent resolutions. That's where larger sensors always have an advantage. And here, medium format would perform better than FF - if MF as a form factor had the amount of money poured into it that FF does, anyway (MF options tend to be limited).

Direct link | Posted on Mar 26, 2015 at 21:48 UTC
In reply to:

AlexisH: One more thing you don't need with a mirrorless camera. Although I guess they could still finely tune auto focus to account for focus shifting.

So you're saying that F2.8 is 'super shallow DOF'? Or that just the center AF point is good enough, even though focus-and-recompose falls apart for fast wide primes?

Are you also saying that manufacturing tolerances and QC are so good today b/c everything is calibrated in the factory?

I'm going to repeat something I find myself saying quite a bit on this site: just b/c it isn't a problem for you, doesn't mean it isn't a problem for everyone else. As is evidenced, in this case, by the thousands of reports of focus problems readily available around the 'net.

Direct link | Posted on Mar 26, 2015 at 08:18 UTC
In reply to:

stevo23: I'm really disappointed in these samples - grainy, noisy, poorly focused. This has to be one of the most useless "real world sample" pages I've ever seen. Do these guys actually know how to use this camera? I make better images with a small sensor point and shoot.

stevo23: "I don't think it has anything to do with the MTF of the lens because I've seen better results already."

You've seen better results from people who've applied sharpening to their photos? After I've already told you we, by principle, don't apply sharpening to our gallery photos? Which, incidentally, is why I referred you to the MTF curves to explain the relative softness of the lens at F1.4.

Please, think on that for a bit before you respond.

Direct link | Posted on Mar 26, 2015 at 08:14 UTC
In reply to:

D Gold: >Users of the 5DS and 5DS R have much higher, stricter standards.

Respectfully, this statement indicates either a lack of understanding of the amateur photographer of today or was mistranslated. I kind of doubt the mistranslation, as that assertion is made more than once.

If anything, a lot of amateurs want at least as much, and perhaps more than many professionals. Just a perspective, but Canon has a rich history of saying, "You want a better image, or autofocus, etc, you are going to pay". Nikon, the same, but a bit better about feature sets and image quality on the lower end.

In the end, I hope the competition from Fuji, Sony, Olympus, etc, continue to put pressure on the big 2.

"Thanks! My usual workflow entails overexposing a bit and then pulling highlights back...6D and other Canons have massive higlight detail imo, even in seemingly overexposed highlights..."

Just for the record - no camera to date, that we know of, has more highlight recoverability than another in Raw, as highlight recoverability is nothing magical at all and, instead, is determined by your choice of exposure.

The reason some people might say Canon have 'reduced DR' is because you can't get shots such as [this one](http://www.dpreview.com/galleries/reviewsamples/photos/3168741/rishi-dsc09922-edit-11mm-f11?inalbum=canon-ef-11-24mm-f4-l-usm-samples-gallery) without using multiple exposures or filters (which, btw, aren't an option with the stellar new 11-24).

Direct link | Posted on Mar 26, 2015 at 08:10 UTC
In reply to:

Robin Ducker: While this is an impressive performance I would be far more impressed if there were some near-field shots with objects taken at less than a metre and extended into 5-10 metres away. This is where a lens performance can really be judged because infinity focus is just one parameter.

Second, buyers of the 14-24 2.8 are attracted to the lens' build quality and durability so It would be good to have some sort of insight into how long the Tamron could perform. Not knocking - just acting like a real lens buyer.

Rick, I hope these sorts of tests were what you were looking for :)

Direct link | Posted on Mar 26, 2015 at 08:02 UTC
On New samples from the Sigma 24mm F1.4 DG HSM Art lens article (235 comments in total)
In reply to:

Storky: The most critical issue with this lens is: does it autofocus correctly? That's why I (and many others) swapped the Sigma 35 1.4 for the Canon original. The Sigma may be somewhat sharper, but it just doesn't focus reliably. The USB dock is an admission of this, and represents a partial, duct-tapish solution. And being slightly off focus at 1.4 is much more detrimental to image quality than small difference in optical corrections.

No, my tests are not like asking Bill if Hillary will win, & you know it. Your polling analogy was about asking someone who they will vote for. The reliability of that poll has absolutely no correlation w/ the knowledge level of the person being polled, & you know that too.

Meanwhile, if a user complains that a lens that needs +14 AFMA 'doesn't focus well' and hasn't calibrated his lens, that's an invalid data point for AF *precision*. If a user tests repeatability on a moving subject, or while he/she is moving, that's an invalid data point. And I've found many, many complaints about native Canon/Nikon sub F2.0 primes not focusing properly... a lot of the time, it was determined that it was simply b/c the user hadn't microadjusted.

Now, AF *accuracy* (the optimal AFMA value) may be worse for 3rd party lenses, as we don't know what bodies they calibrate their lenses to. That I could believe. Perhaps that's what people are actually complaining about; yet it's a largely solvable issue.

Direct link | Posted on Mar 26, 2015 at 08:01 UTC
In reply to:

cinemascope: I am so happy that we have an excellent middle ground system that doesn't go to extremes for marketing purposes and is very good at what it is...
I have nothing against FF but I am so tired of FF evangelists spreading the gospel like it is the best thing since sliced bread.
Even if FF was cheaper than m4/3 I'd still happily pay more for m4/3 because this is what I want...
Like I prefer compact Asian cars as opposed to American guzzlers, or like I'd rather buy organic food even if it costs twice...
I also have been noticing a constant undertone in recent reviews here along the lines of "sure it is not as good as the likes of FF but bla bla" which is getting very tiring... I recall dpreview being a lot more neutral in that regard in the past...

"It kicks in at the same point it would kick for FF if it would attempt the same DOF with the same pixel density."

Yes, that's absolutely true... which is why I said that diffraction is correlated with *equivalent aperture*, not *aperture*.

Do you agree that we can leave resolution out of the conversation if we talk about *normalized* comparisons, which I think is more relevant?

Direct link | Posted on Mar 26, 2015 at 07:55 UTC
In reply to:

AlexisH: One more thing you don't need with a mirrorless camera. Although I guess they could still finely tune auto focus to account for focus shifting.

'No one really needs this' is effectively like saying 'no one really needs F2.8 or F2 or F1.8 or F1.4 full-frame lenses'...

Direct link | Posted on Mar 26, 2015 at 07:23 UTC
In reply to:

Jan Madsen: I purchased the previous version. A good idea, but unfortunately a completely useless system. Far too many bugs, and when I eventually got a measurement / AF correction it was often way off for large aperture lenses - where it really counts. For F/2 and smaller apertures it works ok, for f/1.2 the results fluctuates, and for the Canon 50mm f/1.0 it is almost impossible to get anything meaningful out of the system.

Now if only Reikan would offer support, and correct the numerous bugs, fine, but this company has offered the worst support I have ever encountered with a commercial product - that is none. You can create a support case, and then - nothing. No answer for months. Only way to get any kind of reply from them is to plaster their Facebook wall with notes about the poor service - that worked twice, but then silence again. Unless Reikan has made a dramatic change of attitude towards the customers I would say use other options. Sadly.

Are you saying that for f/1.2 or f/1.0 lenses the results fluctuates when using the software itself, or when shooting in the real-world?

Direct link | Posted on Mar 26, 2015 at 06:55 UTC
Total: 1346, showing: 1 – 20
« First‹ Previous12345Next ›Last »