30 Jul 2012

Canon 1Dx or 5D3 the Best Night Shooting Camera?

For our own time-lapse, stills and video- we wanted to know whether the brand new 1Dx or 5D3 is the best Canon night shooting DSLR. Surprisingly, there’s just about no information around online or elsewhere to clear up this question in regards to High ISO comparison at the extreme levels we use in “Shooting the Stars.” Yes, this is a narrow question considering just how much else the cameras can do- but this one aspect has big implications for our work involving the night sky. We love being out under the stars and want to continue in capturing that experience, so knowing which camera handles High ISO’s the best will directly influence what we bring or rent on our next trip out. Considering the massive price gap between the two models (1Dx $6,800 – 5D3 $3,350) we thought many of you would be curious about this issue also. With some of our gear and help from our local camera store, Pro Photo Supply we got a 1Dx and 5D3 and also the older 5D2 and 1D4 to go out and take actual star pictures to try and settle this. It was important to us to show the 5D2 and 1D4 to give samples of what HAS been possible in night photography compared to what is NOW possible. Apologies to the Nikon crowd, we don’t mean to be discriminating, but we work on the Canon platform, so for this round, we only tested Canon equipment. **Update- looks like we’ll be testing a D4 vs 1Dx later this month**

[Sidenote- all of this equipment can be bought at your local camera store- but, if you want to try out “Shooting Stars” yourself at a very affordable rate and with the newest models we recommend you consider renting it. For any Portlanders- we love  Pro Photo Supply; but for those out of town, Borrowlenses.com has great rates and service. You can find the 1Dx , 5D3 and our test lens- the 14mm f/2.8 L at those links. After three years of shooting stars- I highly recommend using rental services to experiment with top level gear in this tech-heavy and high-priced niche of photography.]

Our idea was to bring all the cameras out to one place and use one lens, one tripod, same settings, and then work through the various ISO levels on each camera. That should provide some information for which body excels at the High ISO’s, or at least be a starting point to get some comparative media. So we got a 14mm f/2.8 L lens and then looked for a place to take some star pics. It just so happened we had weekend plans to hike and camp on the summit of the South Sister, a 10K’ mountain about 3 hours from Portland. What better place to test out night shots than almost 2 miles up in the air? That idea sounded awesome, but proved to be one heck of a butt-kicker lugging up 5 tripods, 8 lenses, 5 DSLR bodies, and a timelapse dolly system. We made it up there, but…. I had to put some of the gear in friends backpacks to get it all up there… so embarrassing.

After much sweat and taking breaks (for me), our group made it to the top. And it was beautiful.

I’ll admit, it was a blast to have our “office” be in such an incredibly inspiring spot. We were geeking out on cameras, being overwhelmed by the sparkling stars above, and standing on edges with bone crushing drops. Add gusting winds and some major energy fatigue- and it makes for a unique and exciting experiment!

Let’s get to it. We set each camera to 25 second exposure, f/3.2, and then worked through the various ISO levels. To manage our time up there, we started at ISO 3200 and then went forward every other ISO level up- 3200, 5000, 8000, 12000, etc. I’ve blocked these images in ISO groupings below, showing all the cameras in one image. In the field, it was difficult to get any sense of whether the 1Dx or 5D3 was taking better pictures. The 1Dx has one higher level of ISO available (ISO 204,500??!), but didn’t have obvious better image quality than the pictures showing up on the back of the 5D3’s. However, both the new cameras crushed the older 5D2 and 1D4. Normally, I don’t go beyond ISO 4000 with the 5D2 or 1D4- usually trying to shoot at ISO 2500 or 3200. With the 1Dx and 5D3, I was confidently shooting at ISO 5000, 8000, and even thought 12,800 looked good. So, initial reaction to the new cameras while shooting is that neither seemed to hold the advantage over each other, but both were significant leaps forward beyond their previous models. Canon Techs, we salute and thank you.

Since I didn’t have an obvious winner in the field between the 1Dx and 5D3, I shot a bunch of test shots to bring back and scrutinize more closely back on the computer. I’ll walk you through the results and to what I think is an apparent winner. But first, we’ll start with images straight out of camera and then work down to images I’ve done controlled edits on in post. The differences really show up when we start ramping exposure and shadow levels in post processing- but we’ll start with no edits first.

This is ISO 3200, Raw to JPEG conversion in Adobe Lightroom 4, with camera names on the corners of the image. I’ll start off by telling you what you’re looking at. Each image is the same since it was shot from the same tripod and lens. In the lower center is the Middle and North Sister, 10,056′ and 10,085′ summits respectively. The orange light on the left comes from Portland and Salem, OR- two of Oregon’s largest cities. Portland is over a 100 miles away and Salem is almost 90 miles away, and yet the light pollution glow is terribly significant. On the right, the much smaller towns of Redmond and Sisters glow on the horizon. Even 2 miles up, in the middle of the mountains- light pollution overwhelms our horizons. ISO 3200 is very high for normal photography, but you can see it barely begins to pick up the light in the night sky, and definitely not the shadow details. The 5D2 in the bottom right doesn’t handle the color gradient of the sky so well, and as the test progresses you’ll see that the 1Dx seems to pick up Aurora activity in its images, try not to let the Northern Lights bias your opinion…

**Note: in all the test images below, you can right click and open a larger res version to see better the details of each image**

Next up is ISO 5000. Aurora purple and pinks in the 1Dx image is more apparent, but still not much visible difference in the the images. The 1D4 is showing some horizontal banding in the sky and the 5D2 is getting “blotchier” in the corners of the sky.

I don’t have an image for ISO 8000 from the 5D2 (fatigue: 1), so we’re jumping up ahead to ISO 12,800. In the past, I would never go this high with the 5D2 or 1D4. Even so, straight out of camera, it’s difficult to detect the noise that is lurking in the shadows, but you can see the image is now starting to achieve some balanced image illumination straight out of camera with the sky and foreground- it’s no longer one big dark shadow. Aurora activity is obvious in the 1Dx and notice now the 1D4 image is actually cropped tighter than all the rest. This is because the 1D4 has a 1.3 crop sensor, while all the others are full frame. This is a good time to point out a negative aspect of the crop sensor when shooting night skies. Are you beginning to notice that in the right side of the sky there is some sort of mottled pattern that is discernible but not really definable? That is the Northern portion of the Milky Way. The iconic Milky Way shots are taken of the Southern portion of the Milky Way (front the Northern Hemisphere…I know, kind of confusing). This section is much more cohesive and “put together.” The Northern half looks like someone took a salt shaker and haphazardly spilled stars in a jagged line up the sky. I point this all out, because the 1D4’s cropped image doesn’t give as much context to the landscape below, but just as impacting, the Milky Way in the sky looks crowded; without more contextual space around it, it’s difficult to tell it is the Milky Way where the full frame images give it space around so you can see where things begin and end. This extra space is a big reason I love the full frame sensor on the 5D2, and now also the 1Dx and 5D3. For shooting stars- full frame is best. Becoming noticeable now is the color noise in the 5D2 outside edges. See the purple-ish color in the shadows at the base of the image? Also notice the sky is mottled between greens and a blue-purple in the top portion. The 5D3 image seems, natively, just a tiny bit brighter than the 1Dx. The 1D4 and 5D2 have a shooting star in the frame.

Here is ISO 20,000. Note: it’s actually 25,600 for the 1D4 and 5D2- they don’t have a 20,000 setting. I’ve been shooting night stars and landscapes for over three years now. When I started on my Canon 30D, I thought pushing to ISO 1600 was crazy. Now, working with cameras operating at ISO 20,000 is mind boggling- and we still aren’t at the cameras limits! Here at this ISO, we’ve achieved some shadow illumination in the darks. Our skies are blowing out in the center- notice the pink/purple Aurora in the 1Dx is barely noticeable in this overexposed state. The 5D2 has apparent color shift in the shadows with shadows incorrectly displayed in purplish hues. Also notice the horizontal and vertical banding in the 5D2. The 1D4 does not show the color shift in the shadows, but keep in mind it is cropping in from the vignetting edges of the lens, which are the first problem areas when shooting wide open. Both the 1Dx and 5D3 show superior contrast in deeper shadows. The 5D3 is just now barely beginning to show some purple color shift in the far lower right corner. Both the 1D4 and 5D2 caught a shooting star in these frames.

For sake of time, I’m going to jump ahead to the last, highest possible ISO on each camera. This is the “H2” (or H3 on the 1D4) in all the cameras. It is not meant to be a “usable” ISO, and so many in the photography community have blasted camera makers asking why put it in there at all? That’s a very fair question. Truthfully, these images are so terribly noisy that there is no recovering them for any professional or reasonable use. HOWEVER, once again, night photography is unlike most other forms of photography and we do things a little different around here than most other styles. I LIKE the H1 and H2 ISO settings because I use them to do test shots when I show up to a location. It’s dark out, I can’t see and I need to setup my shot. Why wait for a “clean” ISO 3200 test shot at 30 seconds when I can crank up to H2 for an 8 second exposure that lets me see into the dark, straighten my horizon, work out my focus issues, improve my composition, etc. By using H2 with a shorter exposure time, I’ve sped up my shooting process therefore getting more shots in a shorter period. For me, this is a good thing. Also, H1 and H2 are a hint of what will be usable in future camera generations. So, quit complaining about the H1 and H2- Camera Techs, we salute and thank you…again. Now let’s talk about it.

WOW, look how bright the night has been rendered! And, notice the ISO levels- 204,800 and 102,400??!?! Holy Blinding ISO Batman!!! Take note that the highest ISO on the 5D2 is a “measly” 25,600, this will influence it’s “underexposed” look compared to the others. At these extreme ISO’s all images are showing the purplish color in the shadow- even the 1Dx and 5D3. Noise artifacts are apparent everywhere, and it’s so bright that the night sky is blown out in the centers. This is not a usable ISO setting, but do notice that it incredibly illuminates the landscape for tilt and focus correction. As I mentioned- I think we’re looking at usable future ISO settings a couple generations away. Mind boggling? Impossible?  Maybe, but that’s what I thought about ISO 4000 back in my Canon 30D days….

I was surprised to have made it this far in the comparison and still not have seen an obvious champion for High ISO use. The 1Dx and 5D3 seemed neck and neck in regards to straight out of camera images. Still hunting, I chose ISO 12,800 and looked at 100% zoomed in views of various parts of the image. I checked out the star fields, how it handled the city lights, and center detail of the mountains and some sky.

Looking at the center of the image with details of the mountains- not really an apparent big difference. The 5D’s seem to have a bit more “snappy” illumination on the whites in the snow fields, where the 1D’s have less contrast. The purple shadows of the 5D2 are very apparent here.

Next, I went over to the city lights of Redmond and Sisters to see if the cameras resolved the brights and darks any better than each other. The 5D3 and 1D4 seem identical, the 5D2 is obviously the worst, and now the 1Dx seems to nose out in front of all the others with a more rich dark shadow below the city. Notice all but the 1Dx have a very noisy shadow image. Also notice how the city lights on the right have a “spiky” look to them? That is lens distortion on the far edges when shot wide open. The 1D4 doesn’t show it because it has cropped in from that part of the lens.

Next, I zoomed into the star fields and looked for a difference there. Shooting stars are burning through all but the 5D3, but really not much of a difference in regards to camera performance.

Alright, we’ve worked through the ISO levels, zoomed into 100% view in the image- and still not really found an apparent winner. But, now it’s time to turn a corner. Up till now, we’ve been working with “straight out of camera” images. In other words, no post processing just convert RAW to JPEG. The corner I speak of is not in-camera Noise Reduction- for this experiment I didn’t use any In-Camera Noise Reduction, and let me tell you why. Any Noise Reduction is ultimately destructive to the integrity of the image. It changes how edges blend, how colors interpret, and how tones are expressed. Yes, ultimately this helps the overall image when using High ISO’s, but I want to control and manage that process as much as I can. So, in the field I want to take the best image I can and bring home the highest quality version of that image for me to then process and manage the noise IN POST under my control. Basically, I’m a control freak. I’m not going to leave it up to the camera to decide how much to permanently alter the pixels through In-Camera Noise Reduction, instead, I maintain the right and ability to decide how much these factors are pushed and pulled. And, by bringing home a non-Noise Reduction image, I have a clean RAW file that can get better Noise Reduction as software increases in the future. So, based purely on my own control-freak obsession, I am not including In-Camera Noise Reduction in this test.

The corner we will turn here is how do the RAW image files hold up to post processing exposure and shadow ramping on the computer at home? From the conception of the test idea, this was the real “litmus” for me personally. As you heard in the beginning- I like to use ISO 3200 mostly, but you also saw that ISO 3200 is very dark and seemingly underexposed. So, in Post Processing I ramp exposure and other levels to bring the image to a desired brightness. I prefer ISO 3200 because it seems to be the sweet spot for getting light from the night scene, but also being able to manage the noise levels in post production. If I go higher to 4000 or 5000, I usually can not manage how much noise is in the image. If I shoot at 2000, I usually am not satisfied with the amount of illumination in the image; an image shot at ISO 3200 with post processing has been my sweet spot. I use Adobe Lightroom 4 for 90% of my editing of night images, so that’s what I used for this test. I’m not going to go deep into my post processing workflow, but rather simply scratch the surface to see if there is any apparent strengths in one camera’s RAW image attributes than the other. Since most of what I do is about brightening the image, particularly bringing exposure to the shadows, I used a simple +2 exposure bump in Lightroom’s editing on all the images as first round of Post testing.

This is the ISO 3200 grouping with a +2 Exp bump in Lightroom 4.

Now we’re starting to see some apparent differences even in the “low” ISO of 3200. The 5D2 is heavily showing the purplish color noise in the shadows and the sky is much more “mottled” up in the corners. The 1D4 has some color breakdown in the gradient of the sky struggling to interpret the green into deeper blues. 1Dx and 5D3 still holding nose to nose.

So, then I decided to really push things. Enough of this delicate tiptoeing around. Let’s crank these RAW files and see what happens!

I cracked the whip over the heads of the horses to get them running and see who would pull out in front. After thinking about it for awhile, I figured the best way to do this is to push the exposure and shadow ramping higher and higher until only one camera was able to keep up. I used the same ISO 3200 images, but applied an unheard of +3.25 exposure, +15 shadow slider and +10 Shadow on the Curves adjustment on the RAW files. All these tweaks are made with the intention to turn the shadows inside out and see what’s lurking underneath the surface and how far can these RAW files be pushed? I’ll tell you ahead of time, I tried the same thing with the highlights- but they all recovered highlights just as well as each other, so predictably- the ability to cleanly up the exposure into the shadows of the High ISO images is the true determining factor for these night shots. In other words- the winning image below became the deciding test of which camera is the best night shooter.

In the image above, I pushed all the cameras until they all began to fall apart except for one, and then I pushed that one as far as I could up till I felt it was at the threshold of “usable.” In my personal opinion, by this test- the 1Dx is the unquestionably better night shooting camera of all the current Canon DSLR’s. As you can see, all three of the others are losing color purity of the shadows, with purple showing up in all but the 1Dx. Horizontal and vertical banding has become terrible in the 5D2, and noise levels in the 1D4 are completely irretrievable and even it is showing purples in the shadows. The 5D3 is on the border of recoverable, exhibiting the purplish shadows but still at an amount that could be managed. Regardless, the amount that the 1Dx seems unfazed and still “in control” of shadow noise, contrast, and color is phenomenal- still with NO Noise Reduction (!!) and is the sole reason I am entitling it Canon’s Best DSLR Night Shooting Camera. The ability to push ISO 3200 this far and have a usable image before any Noise Reduction is very exciting.

Both the 1Dx and 5D3 have come significant steps further in the realm of night photography. To find out how much further, I wondered how much better the images would look if I gave them some significant post processing. In other words, given some spit and polish- which image would shine the best? Below is that comparison using the ISO 3200. I am not listing all the tweaks and adjustments made on each image because this post is not about how to process night images. You’ll just have to accept that I did my best to make each image shine as best it could using the tools in Lightroom 4. Immediately the 1Dx and 5D3 have more contrast and “brilliance” to their images. Comparatively, the 5D2 and 1D4 seem dull and flat and washed with a blue/purple in the whites. Between the 1Dx and 5D3, it’s again difficult to find a winner here. In this scenario we’re not cranking the shadows to their breaking point, so at the “safe” level of ISO 3200 both seem to perform equally well. I think the 1Dx image catches the eye simply because of the Aurora colors in it. At this usage level, there’s not a big difference between the 1Dx and 5D3.

I scratched my head some more trying to think up other ways to put these cameras head to head for ISO usage. I thought maybe there’s a difference in how high of an ISO I could control and use at a level that was acceptable? This proved to be an interesting test. Moving up through each camera individually, I edited it how I usually would, but kept going up through the ISO’s until I was at a level that I couldn’t manage the noise and then went back down to the previous level. Keep in mind, in the field, we shot images at every other ISO level. So, there’s a good chance the next 1/3 ISO stop up may be usable, but wasn’t available for testing. For example, I know from experience that on the 5D2 ISO 4000 is just usable, but I didn’t have that to test. Impressively the 1D4 was manageable up to 5000. The 5D3 was NOT usable at 8000, but I’ve heard 6400 is manageable and I believe it. But unfortunately, I didn’t have one shot at that ISO for testing, so it stopped at 5000.

Amazingly, the 1Dx was controllable all the way up to 12,800! By controllable, I mean a night shot at this ISO- I felt confident in how I edited it that I would post it online as one of my professional portfolio night shots. That doesn’t mean I’d make a print out of it (digital display vs professional print are leap years apart in regards to ISO’s) but I am confident to show it online. To me, this is a big difference and another reason the 1Dx champions out front of the others.

That’s our test and I hope you find it valuable for your curiosity or research in planning your next Canon purchase. This test is solely constructed to give feedback for Night Star shooting, so my apologies in advance for all the other ideas and questions it does not answer or accommodate. As I mentioned in the beginning, the price difference between the 1Dx and 5D3 is very large with the 1Dx costing TWICE as much as the 5D3. In regards to Night Photography alone, I don’t think the 1Dx’s abilities justifies that size price jump. In fact, bang for buck, I think the 5D2 is still the best “entry level choice” since used 5D2’s can be found on Craigslist for under $2k, as opposed to new 5D3’s at $3,500. However, if you want to be on the “front lines” of night shooting, an upgrade to the newer bodies is a must as it is evident their abilities to handle higher ISO’s has expanded considerably and is very exciting. If you have an older 5d2 and are considering a new camera- I think the 5D3 is a fantastic purchase if you are or want to be in Night Shooting. Get the 5D3 and save that other $3k for one or two L series lenses. However, if you do have the budget, and want the best, the 1Dx represents the cutting edge of Night Photography with a Canon camera.

One odd thing I’ve noticed before, now definitely in this test- there’s a big difference in how the 5D series and 1D series interprets shadows. The 5D2 notoriously expresses shadow detail with a purple hue and the 5D3 shows this trend still continues; whereas the 1D4 has to be pushed to its limits before the purple shows up in the shadows. I thought this was mostly due to the crop sensor excluding the lens edges, but the 1Dx- full frame- follows the trend of the 1D4 by not degrading shadows with purple hues. I’ve asked a Canon rep why this is and he’s forwarded it on up to a uber-geeky tech, so hopefully we’ll get an explanation about this. It certainly stands to reason that color-integrity in the shadows is a perk of paying for the higher model, but I’d still like to know what’s the difference. I’ll let you know if we get an answer.

*Laughing* This might be a good time to repeat Borrowlenses.com has great rates and service. You can find the 1Dx , 5D3 and our test lens- the 14mm f/2.8 L at these links. We are looking forward to testing the new Nikon D4 against the Canon 1Dx sometime in the month of August. Stay tuned for that battle of the Juggernauts!

If you allow me to rant further on- I do not think the 1Dx or 5D3 represent a “game changing” jump in night shooting. Years ago the 5D2 and Nikon D90 were literal game changers in their ability to controllably shoot much higher ISO’s than previously possible. Those cameras literally changed what was thought possible. The 1Dx and 5D3 provide substantial, exciting progress, but they won’t turn the world upside down. Looking to the future, if I could get my hands on the next camera advancement- having that NOW would probably blow my mind compared to the 5D2 and 1D4. Saying this, I realize how spoiled we have become with the rate of progress in our cameras. As mentioned, I discovered how to shoot stars on the very old Canon 30D body. With that camera, ISO 1600 was usually my limit and now the 1Dx is making ISO 12,800 possible?!? Our expanding ability to go out into the dark and bring back images is incredibly exciting and I have no doubt in 1-3 years I’ll enthusiastically be doing another test up on a mountain with new cameras and higher ISO’s. In the meantime- try to keep something in mind and a priority- ANY TIME, and I mean any time, you get to go out and spend a night under a deep blanket of stars trying to take pictures- don’t let yourself get so wrapped up in ISO’s, f-stops and images that you miss out on the feeling of actually being there. I’ll reach for a comparison here- my dad loves to fish and his favorite line is, “A bad day of fishing is better than any good day at work.” In the same line of thought, I encourage you to remember that BEING out under the stars is the real experience- not the camera or lens we take a picture with.

Follow us on Facebook to leave comments/reaction about this test and future announcements of more trips, tests, and videos here: https://www.facebook.com/UncagetheSoulProductions

Cheers and Clear Skies,

Ben Canales

Uncage the Soul Productions – dev.offthewallmedia.com/uncagethesoul

and

The Star Trail Photography – www.thestartrail.com