Rishi Sanyal

Rishi Sanyal

DPReview Administrator
Lives in United States Seattle, WA, United States
Works as a Deputy / Technical 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. It's been an incredible opportunity to meld these two interests together here at DPReview!

Comments

Total: 1763, showing: 1 – 20
« First‹ Previous12345Next ›Last »
On Panasonic Lumix DMC-G7 real-world samples article (78 comments in total)
In reply to:

Searching: You can hardly call these high octane samples, quite frankly I'm disappointed and was expecting more, these are quite average in every regard.

I like how people are really afraid to like my comment. I bet they're wondering: 'If I *don't* like it, will he *not* flog Sam? I'm all for flogging but that Canon 11-24mm F4L should primarily be used as a *photographic* tool, you know? So yeah I think I'm going to leave that 'Like' button alone'...

Direct link | Posted on May 28, 2015 at 04:39 UTC
In reply to:

ritholtznue: Hi Rishi and DPR,
Thanks for the quick review. Matching d5500 performance with little trade off between noise vs details is nice. And rebels do 5 FPS which is same as Nikon best crop camera (d7200). Not bad Canon.

Can you please come up with a standard workflow (like exposure latitue and stuido scene) to compare live view and video AF performance. I am thinking, lot of rebel buyers would like to use it for both photo and video (family use). Video performance is also very important in this category.

Also, we haven't really talked about it much in Nikon cameras b/c there just *isn't* usable AF in Live View, and the performance of CDAF for Live View AF is a known quantity. Hence we don't expend much effort on it.

Especially since OVF subject tracking and AF is *so good* on the Nikons (and is what one should be using for continuous AF).

We're actually going to emphasize Live View AF more in the Rebel reviews b/c it's a much more usable way of or tracking subjects in continuous AF (subject tracking is as good as unavailable in OVF shooting).

However, you do have a point, and funny enough - we just last week shot a couple videos showing how appalling CDAF is in video/Live View on the D7200 for that camera's review.

But again, not much point actually expending effort 'testing' it, b/c it's a known quantity. We're not going to suddenly take all other CDAF cameras (a7S, some m4/3 cameras, RX100s, etc.) & show them failing/hunting - b/c we all already know they fail in the same manner.

Direct link | Posted on May 28, 2015 at 00:21 UTC
In reply to:

ritholtznue: Hi Rishi and DPR,
Thanks for the quick review. Matching d5500 performance with little trade off between noise vs details is nice. And rebels do 5 FPS which is same as Nikon best crop camera (d7200). Not bad Canon.

Can you please come up with a standard workflow (like exposure latitue and stuido scene) to compare live view and video AF performance. I am thinking, lot of rebel buyers would like to use it for both photo and video (family use). Video performance is also very important in this category.

Yes, it's the T6s that supports continuous AF in Live View.

Oddly, the T6i is always still continuously tracking/focusing, but when you then press the shutter button to take the shot, the camera requires a re-focus - which ends up causing hunting and a significant lag before the shot is taken. At which point your subject has moved. The T6s just takes the shot.

In other words, the T6i is like the 70D/7D Mark II in this regard, which also requires a refocus, making Servo AF of very limited utility. Although, the 7D Mark II does do its refocus much faster, thanks to Dual-Pixel AF.

I don't know why they do this on the 70D/7D2, b/c clearly the cameras *are* focusing. So your subject *is already in focus*, so requiring the re-focus just kills Servo AF for stills.

My feeling is it's just an oversight. It's almost as if they were of the opinion that dual-pixel AF is only/mostly for video. Which is a shame, b/c I think dual-pixel AF is a game-changer for *stills* even.

Direct link | Posted on May 28, 2015 at 00:15 UTC
On Panasonic Lumix DMC-G7 real-world samples article (78 comments in total)
In reply to:

Searching: You can hardly call these high octane samples, quite frankly I'm disappointed and was expecting more, these are quite average in every regard.

OK. We won't fire him. But we'll flog him. Just for you.

Direct link | Posted on May 27, 2015 at 23:56 UTC
In reply to:

ritholtznue: Hi Rishi and DPR,
Thanks for the quick review. Matching d5500 performance with little trade off between noise vs details is nice. And rebels do 5 FPS which is same as Nikon best crop camera (d7200). Not bad Canon.

Can you please come up with a standard workflow (like exposure latitue and stuido scene) to compare live view and video AF performance. I am thinking, lot of rebel buyers would like to use it for both photo and video (family use). Video performance is also very important in this category.

↑ What he said.

Also, yes we're working on some standard workflows for AF, but it's very difficult. We take video through the OVF and Live View as a way of showing subject tracking, and complement that with real-world tests with 100% crops in rollovers. That's the best we can do for now. There's definitely interest in developing motorized rigs that repeat the same motion, but it's very difficult, and also hard to ensure any such repeated test correlates with real-world performance.

But trust me we're trying. :) We'll definitely take some video of the Rebel focusing in Live View, as its subject tracking is actually usable in this mode (not in OVF shooting, though). Which is important for the demographic using this camera (they probably want the *camera* to worry about focus). I can tell you right now subject tracking is great in Live View (as you'd expect, since it's using the image sensor), but it does hunt significantly more than a dual-pixel camera. T6i doesn't AF-C in Live View.

Direct link | Posted on May 27, 2015 at 20:16 UTC
In reply to:

jrkliny: I am truly uncertain how to interpret this new data. When I look at the comparisons it seems that a Nikon 7200 can be pushed about +5 stops without excessive noise. The T6 sensor can be pushed about +3 stops. A couple of extra stops is clearly an improvement, but I am not sure how valuable this will be in normal use.

The 'free Raw converter' is a red herring. Practically speaking, everyone offers a free Raw converter.

That you shouldn't use to begin with. :)

Direct link | Posted on May 27, 2015 at 18:43 UTC
In reply to:

nananananana: I wish people would stop assuming that dynamic range is a defining factor when people buy these cameras. You do realize that you can't see high dynamic range from a regular photographs. The only way Dynamic Range might matter is for people who do very heavy post editing.

We dun care about dynamic range. Stop assuming we do. What I care about is:

-COLOR ACCURACY (thank you low pass filter)
-LOW WEIGHT
-CONVENIENCE
-GOOD JPEG QUALITY
-GOOD APS-C LENS SELECTION THAT ISN'T TOO EXPENSIVE AND LIGHTWEIGHT
-FREE SOFTWARE LIKE DPP
-ERGONOMICS

Canon provides all the above, which is why they ship cameras.

We do not care about dynamic range or micro adjust. And by we, I mean the target audience for these cameras. People who buy these cameras have a life outside of photography, they are not pro photographers.

I know what false color is. That's different from 'accurate colors'. No, it doesn't help you get more accurate colors. Where did you get this idea?

'DPP is much more than a simple Raw converter'? Not any more so than the free Raw converters that Nikon and Sony provide.

That covers probably 99% of the entire camera market, that offer free Raw converters.

But that's an advantage of Canon?

Direct link | Posted on May 27, 2015 at 08:15 UTC
In reply to:

nananananana: I wish people would stop assuming that dynamic range is a defining factor when people buy these cameras. You do realize that you can't see high dynamic range from a regular photographs. The only way Dynamic Range might matter is for people who do very heavy post editing.

We dun care about dynamic range. Stop assuming we do. What I care about is:

-COLOR ACCURACY (thank you low pass filter)
-LOW WEIGHT
-CONVENIENCE
-GOOD JPEG QUALITY
-GOOD APS-C LENS SELECTION THAT ISN'T TOO EXPENSIVE AND LIGHTWEIGHT
-FREE SOFTWARE LIKE DPP
-ERGONOMICS

Canon provides all the above, which is why they ship cameras.

We do not care about dynamic range or micro adjust. And by we, I mean the target audience for these cameras. People who buy these cameras have a life outside of photography, they are not pro photographers.

"We dun care about dynamic range. Stop assuming we do. What I care about is:"

That's great, thank you for sharing. Would you like us to now change our entire website/testing to cater to *only* your desires?

"-FREE SOFTWARE LIKE DPP"

Most cameras come with free Raw converters that emulate most in-camera conversion/features.

'-COLOR ACCURACY (thank you low pass filter)'

Huh? What does the OLPF have to do with color accuracy? Also, your beloved DXO tests color accuracy. I'm sure you think their color accuracy test is biased, though, b/c it tends to be lower for Canons.

Out of curiosity - why do you find it necessary to defend a brand and start threads to try & discredit anyone (DXO) that publishes anything that ends up showing a less-than-ideal score for some aspect of that brand's cameras? You do realize that that same site is soon going to start showing Canon lenses as having the highest resolutions when paired with a 5DS, right? Will they still be biased then?

Direct link | Posted on May 27, 2015 at 02:55 UTC
In reply to:

jrkliny: I am truly uncertain how to interpret this new data. When I look at the comparisons it seems that a Nikon 7200 can be pushed about +5 stops without excessive noise. The T6 sensor can be pushed about +3 stops. A couple of extra stops is clearly an improvement, but I am not sure how valuable this will be in normal use.

Well, ADL is 'active' whereas DRO is more 'passive' in that ADL actually changes focal plane exposure whereas DRO never touches exposure (so you have to yourself dial in negative exposure compensation).

The shadow lifting (tonemapping) algorithms for ADL and DRO may be the same, I don't really know. I think Apical does the ADL algorithm, but not sure who does DRO (Sony itself maybe?).

While ADL is quite nice, I do find it sometimes leads to halo-ing. But on a D4S, it can get you back a rather whopping 1.7 EV of highlight range, which is quite nice. Nikon's free Raw converter can also apply a lot of shadow lifting using high ADL settings in post-processing. ADL tends to preserve more local contrast while reducing global contrast, which makes it a little more usable than 'Flat' picture control (although 'Flat' can make a nice starting point from where you can re-introduce contrast by, say, adding some black clipping or what have you).

Yes, odd Sony called it ISO 3200. Dunno why.

Direct link | Posted on May 27, 2015 at 02:32 UTC
On Video Overview: Fujifilm X-T10 article (119 comments in total)
In reply to:

Neal Hood: Samples definitely seem to have a "darkish" trait to them. I would not want to buy that camera without using it for a weekend or so.

Are you guys maybe talking about the photos that were purposely underexposed to then have their shadows recovered? Apologies they seem out of order - the brightened shot (Raw conversion) should be right after the darker JPEG, but they're out of order in the gallery.

Direct link | Posted on May 27, 2015 at 02:25 UTC
In reply to:

jrkliny: I am truly uncertain how to interpret this new data. When I look at the comparisons it seems that a Nikon 7200 can be pushed about +5 stops without excessive noise. The T6 sensor can be pushed about +3 stops. A couple of extra stops is clearly an improvement, but I am not sure how valuable this will be in normal use.

Yes, I would say high base ISO dynamic range is useful, if not necessary, for such DR modes (ADL, DRO, 'flat' or S-Log2 picture profiles, etc.) to yield good images. Since these modes raise shadows (with ADL purposefully underexposing in order to retain highlights), cameras that add noise to shadows, b/c they have lower dynamic range, will be limited in their ability to provide good results with such modes.

One has to wonder if the lower base ISO dynamic range of Canon DSLRs is a reason for HTP (Highlight Tone Priority) being limited to only 1 EV underexposure and then selective boost. A Nikon-esque 'flat' picture profile would run the risk of exposing noisy shadows.

Interestingly, even the a7S' S-Log2 mode raised the minimal hardware amplification to ~ISO 400-ish levels. I have a pet theory this is b/c ISO 100 shadows on an a7S run the risk of having some noise, b/c of the somewhat lower DR compared to the a7R (but still significantly higher than Canon, mind you). I may be wrong.

Direct link | Posted on May 26, 2015 at 22:14 UTC
In reply to:

jrkliny: I am truly uncertain how to interpret this new data. When I look at the comparisons it seems that a Nikon 7200 can be pushed about +5 stops without excessive noise. The T6 sensor can be pushed about +3 stops. A couple of extra stops is clearly an improvement, but I am not sure how valuable this will be in normal use.

The 'exposure latitude' is useful on a number of fronts - but really only if you shoot Raw. If you do, then extra base ISO dynamic range allows you to lower your exposure to keep highlights from clipping, & then brighten shadows selectively in post-processing to make them visible again. Or rescue missed exposures (for example, if your camera underexposes a backlit scene, or you wish to correct large amounts of vignetting, etc.). You can read more about noise and how it affects dynamic range [here](http://bit.ly/readnoise).

Does it matter for most Rebel shooters? Maybe not, if they're just shooting JPEG in Auto mode. But we don't want to make assumptions about the user of any camera (a pro might buy a lower end camera as a 2nd body, for example), so we try and put all cameras through our standard battery of tests.

2nd body for a pro, btw, is also a reason microadjust should be included. Not to mention a given camera should be able to focus a given lens w/out having to send to factory.

Direct link | Posted on May 26, 2015 at 20:16 UTC
On What is equivalence and why should I care? article (2129 comments in total)
In reply to:

ABuck289: I'd like to see where the medium format cameras fit in. Is a Pentax 645 worth the additional cost?

Great article, BTW.

Well, if you're going for the shallowest depth-of-field, full-frame might be your 'sweet spot', simply b/c - for all practical purposes - full-frame is the only system that offers f/1.4 *equivalent* lenses.

Now, yes, there are exceptions - you could put a rare f/1.0 lens on APS-C to get f/1.4 equivalent, or use a Metabones Speedbooster on APS-C with a f/1.4 lens to get f/1.4 equivalent DOF.

But when it comes to medium format, I'm not aware of any f/1.4 lenses off the top of my head, so please let me know if I'm wrong.

Take a Pentax 55mm f/2.8 lens for the 645Z system, for example. Its crop factor is .79x, which makes it ~44mm equivalent in terms of full-frame (FF). If you do an equivalent aperture calculation, it comes out to f/2.2 (down from the original f/2.8 of the lens).

So you're essentially getting 44mm F2.2 *equivalent* on the 645Z. Point being, you can get to much shallower DOF equivalents (f/1.2 anyone?) on a FF system, b/c of the vast diversity of lenses for FF systems.

Direct link | Posted on May 25, 2015 at 08:45 UTC
On Nikon D5500 Review preview (342 comments in total)
In reply to:

Robert Eckerlin: Jeff

Thank You for the following youseful remark: " One issue I usually encounter with Nikon DSLRs is their tendency to overexpose by about 1/3-stop. What I ended up doing with the D5500 is dropping the exposure compensation by that amount and then bracketing, which worked fine in most situations. "

I too Encounter with my D5500 the overexposure problem (by the way, it was a Problem, that i did not Encounter with my beloved but old D5000).

Can I please ask you a question: When bracketing with the D5500. how often is it, that it is not the photo shot with -1/3 exposure, which you think is the best exposed? Around one 10th of the times? More? Or less?

I will also appreciate feedback from others.

Thank You very much in advance

Pretty sure Jeff was commenting that the camera *in general* overexposes by a bit. You should be able to fix this by biasing the meter, but honestly I can't remember if the D5500 has this option. The higher end cameras do.

Not sure the 'Brightness' option of 'Picture Control' biases the actual meter...

Then again, maybe the overexposed feel of the images is not b/c the actual exposure was too high, but b/c the JPEG engine brightened the image too much - in which case it might be as you say the 'Brightness' option of 'Picture Control'.

Hmm... this is why we should all shoot Raw and get Raw histograms in-camera. Just kidding, of course :)

Direct link | Posted on May 25, 2015 at 05:55 UTC
On Canon EOS 5DS / SR First Impressions Review preview (2981 comments in total)
In reply to:

Discovery Of Light: I really feel like the whole shoot for the highlights, boost the shadows in post is growing tiresome and making us all a little lazy. Yes a little sparingly is fine but it gets in the realm of unnatural pretty quick when you push the shadow slider too far and that's with all cams. I know it's just my opinion but I would rather have more highlight headroom and give my shadows more exposure. That way shot noise IS reduced because you gave the sensor more light and less amplitude. Have I gone mad for thinking this way?

I'm saying I think that most data is mapped linearly to Raw. We've seen no evidence to indicate otherwise... yet. Would be nice though if the relationship could eventually be changed.

Direct link | Posted on May 24, 2015 at 22:48 UTC
On Canon EOS 5DS / SR First Impressions Review preview (2981 comments in total)
In reply to:

57even: I think I am right in thinking that larger photosites generally have more read noise and less shot noise. They will look better in mid-tones, but if you push the shadows in post, noise will show up.

This would normally mean that the 5DR would be better than the 6D, if it were not for the noisy ADC that Canon's use. However, high ISO (ISO variant) shots indicate that upstream read noise is actually really low. What's messing up the low ISO pushed shots is the downstream read noise.

In other words, the DR problem is only an issue at low ISO, and the reason is the same as for all Canons, namely they don't use dedicated column ADCs like Sony do on the smaller pixel chips (D750 and D810).

5DS ISO cap due to ADC? I'd bet it's nothing more than product differentiation...

Especially b/c there's nothing stopping them from doing it *at the very least* in software just to get you a usable JPEG. And at those high levels of amplification, your data's already well above the downstream read noise floor, so there'll be very little difference between hardware vs. software amplification.

I may be wrong, but I can't help but feel it was deliberate differentiation.

Direct link | Posted on May 24, 2015 at 22:46 UTC
On Canon EOS 5DS / SR First Impressions Review preview (2981 comments in total)
In reply to:

57even: I think I am right in thinking that larger photosites generally have more read noise and less shot noise. They will look better in mid-tones, but if you push the shadows in post, noise will show up.

This would normally mean that the 5DR would be better than the 6D, if it were not for the noisy ADC that Canon's use. However, high ISO (ISO variant) shots indicate that upstream read noise is actually really low. What's messing up the low ISO pushed shots is the downstream read noise.

In other words, the DR problem is only an issue at low ISO, and the reason is the same as for all Canons, namely they don't use dedicated column ADCs like Sony do on the smaller pixel chips (D750 and D810).

Yes, the prevailing hypothesis is that the a7S has some relatively upstream 'switch' to higher gain at higher ISOs, which helps it overcome the non-zero sources of noise in even Exmor architectures. Bill Claff's data shows this as kinks in the DR curves. One at ISO 1600, which perhaps is where the higher (photosite? sensor-level?) gain occurs. Then again at ISO 102k, which I'd bet is noise reduction.

DxO's DR data *kind of* shows it happening, but not as clearly/dramatically as Bill Claff's data. Odd. Also, SNR 18% values over at DXO don't seem affected by this, which perhaps make sense since small changes in upstream read noise likely mostly affects darker tones, not SNR 18% (well, until you get to like ISO 409k where 18% is made of signals in the singles of photons :)).

But, yeah hard to make assumptions. But generally the photon transfer extrapolations give you reliable indicators of performance.

Direct link | Posted on May 24, 2015 at 22:45 UTC
On Canon EOS 5DS / SR First Impressions Review preview (2981 comments in total)
In reply to:

Mr Ed M: When normalized to print resolutions the dynamic range is competitive with other cameras.

It took a while to load, but, ISO 100 +4 EV looks fine when set to print.

ISO 100 +5 EV should clean up fine at print resolutions with a little post processing.

I don't see any issue with the available dynamic range.

Mr Ed M: Yes, all self-evident, correct, and completely unrelated to my point.

Direct link | Posted on May 24, 2015 at 21:32 UTC
In reply to:

lacikuss: If you alredy own canon gear I'd recommend save a litle bit more money and get the FF 6d at $1,399 Body only in Adorama.

If you're looking into APS-C sensored camera without any past equipment then the Sony A6000 is the best overall for the price.

Dynamic range *is* correlated w/ sensor size, though there are other things at play that, for example, make Canon FF cameras fall behind smaller sensor peers (namely, downstream read noise, which you can read about [here](http://bit.ly/readnoise)).

But it's not about smaller pixels blowing out more easily. Remember that for the same exposure, those smaller pixels see less light anyway, so their chances of blowing out are the same as on a larger sensor. It's more that the collective shot noise contributions are higher w/ a smaller sensor, which you can also read about [here](http://bit.ly/shotnoise), which ultimately limits dynamic range of even a perfect sensor.

"An APS-C camera has more DOF and definitely is an advantage for macro work where DOF is a major factor. "

It doesn't have a DOF advantage, b/c you can stop the FF down. Unless you're talking about f/22 on APS-C, in which case, you're right, lots of FF lenses don't let you go beyond f/22. But now you're fighting diffraction.

Direct link | Posted on May 24, 2015 at 19:34 UTC
On Canon EOS 5DS / SR First Impressions Review preview (2981 comments in total)
In reply to:

Mr Ed M: When normalized to print resolutions the dynamic range is competitive with other cameras.

It took a while to load, but, ISO 100 +4 EV looks fine when set to print.

ISO 100 +5 EV should clean up fine at print resolutions with a little post processing.

I don't see any issue with the available dynamic range.

No, not directed at you.

Thanks for sharing your experience - it's a valid point that in many situations, you won't need the extra DR. I'd never argue against that.

Yet I have situations that exhibit the opposite: shooting a horse galloping on the beach at sunset (I've shared this elsewhere - also in my gallery) where the high ISO shot blew the skies, but ETTR'ing via lowering the ISO saved me the skies, and meanwhile didn't introduce camera/electronic noise when I pushed shadows. Same w/ weddings w/ colorful uplights - high DR cameras mean these don't blow/band b/c I underexpose.

And again, I'm tired of making this point, but - it's not just about 4-5 EV shadow pushes. It's also about 2-3 EV pushes that *don't add noise*. Meanwhile, even moderate 2-3EV pushes of cameras with high downstream read noise lead to noisier shadows/midtones than a camera with low downstream read noise (high base ISO DR).

You can see this for yourself in our Exposure Latitude and ISO-invariance widgets.

Direct link | Posted on May 24, 2015 at 19:05 UTC
Total: 1763, showing: 1 – 20
« First‹ Previous12345Next ›Last »