50mm SHOOTOUT! Detailed Analysis of the Hand-Holdability and General Sharpness of the RF50/1.8 vs. the RF24-105/4LIS

[SwissFrank]

See my other posts for details of the goals and methodology. But in short, these targets are 1:1 selections from laser printer-printed targets under household incandescent, hand-held in a seated position. The selections are sorted, for each shutter speed, from sharpest to least sharp. A benchmark example starts and ends each row for ease of comparison. The scores are for guidance only; please let your eyes be the judge. Finally, this target is 55lp/mm and the lines are only 2 pixels thick on an R5. If you are reducing 4:1 or more (e.g., width is 2048 or fewer pixels, even the sharpest possible image would become a grey circle, so even the worst of these images or probably still very usable.

135mm SHOOTOUT! Detailed Analysis of the Hand-Holdability and General Sharpness of the RF135/1.8, the RF100-500/4.5-7.1, and the EF135/2

Reikan Focal looks interesting... I'll bookmark it. What's AFMA? From my 90s lens collection, sharpest aperture always seemed to be two stops down. But now, the 135/1.8 and 100-500 (at least at 100-135) are so sharp wide-open that I think a 45MP sensor can't even make use of any more...

www.canonrumors.com

100mm SHOOTOUT! Detailed Analysis of the Hand-Holdability and General Sharpness of the RF100/2.8 Macro, the RF100-500/4.5-7.1 and the RF24-105/4

See my 135mm shootout thread for details on the methodology, but I'll summarize here: These are 55 lp/mm charts, which make the black lines about 2 pixels wide on an R5. This target has detail that is totally invisible unless your final image is reduced less than 4:1, and really only if...

www.canonrumors.com

CONCLUSIONS:

1. Stopping down both of these lenses seems to bring a lot more sharpness than long shutter speeds cost. The best images in the series were the point I stopped the 50/1.8 down to f/2.5, even though the shutter speed was 1/4 sec! Even the worst of ten shots at 1/4 was excellent. For the zoom, there was a marked up-tick when I stopped down to f/5, even though it came when I was also going from 1/2 sec to 1 sec.

2. For the 50mm, the worst shot from 1/2 to 1/30 sec was sharper than the sharpest shot at any speed above 1/30! So even if you have a huge amount of light, try not to shoot this lens faster than 1/30, unless you need to stop action or have too much light. Go down to 1/2 sec with full confidence.

3. The 24-105 has unreliable IS until you get at 1/30 or better. Then it's reliable. But here, 1/30 to 1/60 is your sweet spot, unless you need to stop action or simply have too much light.

4. If you can shoot 1/30 or faster, there's no reason to swap lenses. But slower than that, the 50/1.8 does a more RELIABLE job than the zoom. This is counterintuitive as the 24-105 has in-lens IS while the 50/1.8 doesn't. Further, you can (should!) stop the 50/1.8 down at which point it becomes sharper yet.

 

[SwissFrank]

Which is better, remote release? 2-second timer? Or single-shot mode?
No real pattern between remote (R) or timer (2) but both are likely to give somewhat better results than single-shot (S). No surprise here. But it's interesting to see how much better. (Ignore the scores except as guidance; compare with your eyes.)

If you have time to work from a tripod, which aperture gives sharpest results with the 50/1.8?
f/2.5 is a good bet, though f/5 and f/5.6 also did well in this shoot. Does anyone have any explanation why there's a valley in sharpness around f/4.0?

Methodology: Gitzo G1228 carbon fiber "Mountaineer" tripod with an Arca Swiss B1 ball head. Two samples were shot with each method, at each aperture from f/1.8 to f/5.6.

Note: don't compare scores between charts; the ranking is based on unrepeatable effects such as exact distance to target. Snippets are sorted from sharpest to least sharp, with a benchmark shot at the beginning and end of each row for sake of comparison to what's "optimal". As always, target is laser printed in domestic lighting, and about 55 lp/mm on the sensor, making each line about 2 pixels tall on an R5 sensor. To look at this chart 1:1 is to be looking at the R5 images 1:1. Files were the smooth L jpg, though they were all the same so I doubt it matters.

 

[koenkooi]

Is that with MS, EFCS or ES?

 

[SwissFrank]

Is that with MS, EFCS or ES?

Thanks for reading and yes, good question. Electronic Shutter! That is why, past 0.5", I started increasing ISO instead of using a longer shutter. By the way do you know why the R5 only goes to 1/2 second on Electronic Shutter?

I plan on doing the same kind of comparison between Manual Shutter and Electronic Shutter.

 

[koenkooi]

[…] By the way do you know why the R5 only goes to 1/2 second on Electronic Shutter?[…]

I don’t know why and many people have used that restriction to keep mentioning the Canon hammer.

I can’t really think of a technical reason for it, it might be the typical ‘If you care about quality, use full-bitdepth mechanical shutter’ thinking.

 

[SwissFrank]

it might be the typical ‘If you care about quality, use full-bitdepth mechanical shutter’ thinking.

Is electronic shutter less than full-bit depth? I don't see any description of quality differences in the Advanced User Guide for firmware 1.5.0. It talks about light streaks from flickering lights, rolling shutter and so on, and lots of limitations about other modes it can be combined with, but nothing specifically about quality...

 

[koenkooi]

Is electronic shutter less than full-bit depth? I don't see any description of quality differences in the Advanced User Guide for firmware 1.5.0. It talks about light streaks from flickering lights, rolling shutter and so on, and lots of limitations about other modes it can be combined with, but nothing specifically about quality...

On the R5:
ES: 12-bit
EFCS/MS: 13-bit in H+, 14-bit in H and slower modes.

As far as I know, the R3 is the only Canon MILC body that allows 14-bit capture when using a fully electronic shuttter.

The manual should have this info in one of the last sections that has all the specs, you can also find a confusing summary here: https://www.canon.co.uk/cameras/eos-r5/specifications/

 

[koenkooi]

I browsed cam.start.canon and it looks the latest versions of the manuals stopped listing bit depth drop for various modes I had noticed that the R8 doesn't list them, but I hadn't realized they had been removed from the R5 manual.

 

[neuroanatomist]

I don’t know why and many people have used that restriction to keep mentioning the Canon hammer.

I can’t really think of a technical reason for it, it might be the typical ‘If you care about quality, use full-bitdepth mechanical shutter’ thinking.

Probably, since the R3 in full electronic shutter allows up to 30 s and bulb exposures.

 

[neuroanatomist]

I browsed cam.start.canon and it looks the latest versions of the manuals stopped listing bit depth drop for various modes I had noticed that the R8 doesn't list them, but I hadn't realized they had been removed from the R5 manual.

Odd! It’s still in the search index (i.e., searching ‘bit’ in the manual pulls up the drive mode page where that info used to reside.

 

[SwissFrank]

I browsed cam.start.canon and it looks the latest versions of the manuals stopped listing bit depth drop for various modes I had noticed that the R8 doesn't list them, but I hadn't realized they had been removed from the R5 manual.

Yes I've been looking at: AdvancedUserGuide-c003.pdf, which is R5 Firmware 1.5.0 and was surprised to see no mention even of bit, bit rate etc. I looked at every occurrence of "electronic shutter" in the manual and likewise no mention. All that said, I'm just using the smooth L JPG file, and I think it probably has no effect on the actual results since all shots in a series have the same file type.

 

[Scottboarding]

For the 50mm, the worst shot from 1/2 to 1/30 sec was sharper than the sharpest shot at any speed above 1/30

Maybe I’m misunderstanding this but are you saying that a photo taken at 1/500th of a second would be worse than one taken at 1/2 second? Or is there a dip in performance at 1/60th or something like that?

 

[SwissFrank]

Maybe I’m misunderstanding this but are you saying that a photo taken at 1/500th of a second would be worse than one taken at 1/2 second? Or is there a dip in performance at 1/60th or something like that?

Yes, as strange as it sounds. Look at the second chart in the first post and only use the scores as a guide to what to compare by eye.

But: on the 50/1.8, even the worst of ten shots at 1/2 shot is better than ANY shot at 1/60 or above.

I have NO idea why. It's not quite an apples-to-apples comparison as the ISO and aperture are slightly different. However, I used f/1.8 and got great results at 1/30. In other tests with this software (all on this forum with SHOOTOUT in the title) I used ISO's up to 5000+ before I started seeing worse results with high ISO, so the fact this is ISO 800 is not the problem either.

 

[Scottboarding]

Yes, as strange as it sounds. Look at the second chart in the first post and only use the scores as a guide to what to compare by eye.

But: on the 50/1.8, even the worst of ten shots at 1/2 shot is better than ANY shot at 1/60 or above.

I have NO idea why. It's not quite an apples-to-apples comparison as the ISO and aperture are slightly different. However, I used f/1.8 and got great results at 1/30. In other tests with this software (all on this forum with SHOOTOUT in the title) I used ISO's up to 5000+ before I started seeing worse results with high ISO, so the fact this is ISO 800 is not the problem either.

That’s super weird. Looking closer at the chart you’re completely right. 1/15th being better than 1/500th sounded so bizarre that I assumed I misinterpreted what you said but nope I understood correctly. Thanks for doing these tests, really helpful!

 

[SwissFrank]

Looking closer at the chart you’re completely right. 1/15th being better than 1/500th sounded so bizarre that I assumed I misinterpreted what you said but nope I understood correctly.

To be clear it's even weirder: the worst of ten shots at 1/2 are better than the best of ANY shots over 1/30. Same for 1/4, 1/8, 1/15, and 1/30. I have NO IDEA why.

And the 24-105 peaks at 1/30-1/60, pretty clearly. Use faster or slower when you need to, of course, but if you don't need to, that is your best speed range. I wish the R5 could take a custom program-exposure program specific to each lens. (Maybe it can, I have a hard time reading the manual.)

 

[jaygould]

Thanks for doing these tests. So overall, which lens has better image quality?

In my experience the 50mm STM is both sharper and produces more pleasing images than the 24-105@50, when both are stopped down to f8.

 

[SwissFrank]

Thanks for doing these tests. So overall, which lens has better image quality?

Thanks to all you guys for taking a look at my little project. Basically all my test data is here. I'm only comparing center sharpness and hand-holdability, really.

My takeaway isn't really that lens A is great and lens B sucks. Instead, it's that to get the best out of lens A, shoot it at 1/2-1/15 when hand-holding, if subject motion allows, and trust the IBIS. When using Lens B, remember the sweet spot is 1/30-1/60 despite having in-lens IS as well.

The lenses are different enough you'll rarely need to choose between them on sharpness. Especially if you can easily shoot 1/30-1/60 without super-high ISO, I don't think there's too much difference. Instead, you'll compare on bokeh, hand-holdability, portability, versus wide-angle/tele capability most of the time I think. I'd guess image quality is like the FIFTH most important factor in your decision.

Just look at the test targets yourself. Use the numeric scores as a guide but look mainly at the actual images. Remember these are at full camera resolution. The black lines (and white lines) are only 2.1 pixels wide, so at say 2100x1400 even the sharpest circle here is just a gray circle.

 

[Sporgon]

@SwissFrank This must have something to to with the effects of IBIS at higher shutter speeds ? Did you run a control with camera on a solid, studio grade tripod and no IS or IBIS ?

 

[SwissFrank]

@SwissFrank This must have something to to with the effects of IBIS at higher shutter speeds ? Did you run a control with camera on a solid, studio grade tripod and no IS or IBIS ?

Hey Sporgon, thanks for taking a look. My second post in this thread with images is indeed a TRIPOD test meant to show 1) sharpness at various large apertures and 2) sharpness using shutter button, 2-second self-timer, and remote (S 2 and R).

I'm an engineer and curious about how the results might be coming about. That said, this is just my #5 hobby for me at this point and I simply can't devote much time to it. I don't have any theory worth sharing other than that the IS/IBIS probably uses some heuristics that work better in some conditions than others. Whatever the root cause is, it'd be intellectually satisfying to know, and yet probably wouldn't help us take better pictures.

My main goal was PRACTICAL takeaways, such as: if I can pick a shutter or aperture, what is best? (shutter 1/2-1/15 works great; f/2.2-f/5.6 are pretty much the same but 1.8 is quite good too). How much improvement does a tripod bring and how best to trigger the camera? (notable but small improvement, and 2-sec timer or remote release also a notable but small improvement over using the shutter button).

 

[Fuzzy]

I don’t know why and many people have used that restriction to keep mentioning the Canon hammer.

I can’t really think of a technical reason for it, it might be the typical ‘If you care about quality, use full-bitdepth mechanical shutter’ thinking.

I need a much longer electronic shutter speed too! 30 seconds please!