Ultimate High ISO Review: Canon 5D Mark IV vs Nikon D850 vs Sony a7R III

Ultimate High ISO Review: Canon 5D Mark IV vs Nikon D850 vs Sony a7R III

Some of you may recall spending hours upon hours in a darkroom. The swirling paper and baths of aromatic chemicals – with a reward of watching prints magically appear under safe, red light. There is a similar awe when shooting 30 second exposures in the pitch dark. Seeing the center of our galaxy appear on the back of a digital camera’s LCD screen is similarly magical. In this post, we’ll explore how well that magic is captured using very high ISO on select recent camera models.

high-iso-test-canon-5d-mark-iv-vs-sony-a7r-III-vs-nikon-d850

There are now many excellent cameras for night shooters. Shooters seek out the darkest skies with interesting foregrounds. Others trek to mountaintops and remote coastlines on moonless nights. Some prefer roaming cityscapes for moody drama. Intrepid explorers of darkness push the limits of gear. The “big three” manufacturers – Canon, Nikon, and Sony – offer flagship prosumer models that excel in low light. But which one performs the best at the outer limits of ISO sensitivity?

Preparing for High ISO Testing

I aimed to shed some light on this question and did an extreme low-light “shoot off” between the Canon 5D Mark IV, the Nikon D850, and the Sony a7R III. Tests happened around Truckee, California and on the beaches of Lāna’i in Hawai’i. I shot both stills and video at extreme ISOs. The cameras were on identical tripods and placed side-by-side. Mounting was consistent as were the lenses (as much as possible). Triggers allowed me to get (nearly) simultaneous shots.

I made every effort to keep the tests as “scientific” as I could. I tried to control for exposure time (~30 seconds), aperture (f/2.8), and only varying the ISO in full stops up to 25600 on each camera. Stills are in RAW, with Long Exposure Noise Reduction (LENR) turned both “on” and “off” for various tests. Note that Sony does not allow high-ISO noise reduction when shooting in RAW.

Below are the overall test parameters, starting with just stills. Video will come in a second post. You’ll find links throughout this post to download the RAW files. You may collect these files for personal use, process them, and examine the results.

canon-5d-mark-iv-vs-sony-a7r-III-vs-nikon-d850

Gear and File Settings

The following gear was used:

Canon 5D Mark IV with a Canon 16-35mm f/2.8L (Mark I) lens (see Mark II and Mark III for rent)
Nikon D850 with a Nikon 17-35mm f/2.8D lens
Sony a7R III with a Sony 16-35mm f/2.8 GM lens

Cameras were on Really Right Stuff tripods and ballheads. They were immediately adjacent to each other. Each camera captures as identical a scene as possible. For the Nikon, I had to use the Nikon 17-35mm f/2.8D lens (the Nikon 16-35mm only has a maximum aperture of f/4). This means the Nikon was not going to be quite as wide as the others. So I zoomed the Sony and Canon lenses to 17mm (or very close to it) for consistency.

Cameras were set to Adobe RGB, Auto White Balance, and uncompressed RAW. I have both processed and unprocessed (straight out of camera, or SOOC) images throughout. All content was imported to Lightroom Desktop Classic CC v. 7.3.1 and developed/processed with Camera RAW version 10.3. White balance was made consistent using the eyedropper tool on the center of the Milky Way. Each image ended up being around 3600ºK with -5 tint. Images need to load quickly on blogs, so what you see here is web optimized. Please download the provided RAW files for more accurate visuals.

Shooting Conditions

Clouds and wind stymie any night shooter’s efforts. I battled both of these to varying degrees throughout the tests in both Hawai’i and Truckee. Image similarity and sharpness were occasionally degraded. There are a few times where I could not complete as thorough a test as I’d have liked. The length of time needed to run through five steps of ISO with 30 second exposures in a real-world outdoor application was sometimes cut short by weather changes. Disclaimers aside, I think there are enough RAW images here to help you make an informed decision.

High ISO Still Image Comparison and Results

I shot the first test on Hulopo’e Beach in Lāna’i. Lights from the Manele Bay resort provided some nice backlighting to the foreground. The skies to the east/southeast are very dark, with light pollution from Maui nicely hidden by the rocks of Pu’u Pehe. I shot images of the galactic core at the two highest ISO values of 12800 and 25600.

test-one-high-iso-camera-comparison-review

These were shot in RAW with Long Exposure Noise Reduction (LENR) set to ON and High ISO Noise Reduction (High ISO NR) also set to ON for the Nikon and Canon cameras. Cloud cover came in before I had a chance to test out ISOs 3200 and 6400.

As expected, the Sony and the Nikon had the advantage with their larger files and larger megapixel count. The shadow noise was highest in the Canon files, especially when the files were pushed in Lightroom. Check out the RAW files here:

Hulopoe-Beach-RAWS.zip (363.30 MB)

For the second test, I shot the hotel. It had very bright flood lights and Tiki torches that threw a lot of light onto the beach. There was still some dusk light in the sky. The sand and ocean were front-lit and shadowy.

Manele-Bay-High-ISO-Review-Camera-Comparison

Check out the RAW files from this test here:

Manele-Bay-Hotel-RAWS.zip (331.4 MB)

Be sure to develop them in Lightroom (LR), or a similar program. Push the shadows and bring down the highlights.

Light Painting Test

For this test, I had some friends’ kids use their headlamps to light paint hearts under the cloudy skies while the Milky Way was rising. The goal was to have a very wide dynamic range to allow for investigation into how the cameras’ files responded to bringing down the highlights.

Hulopoe-Beach-Light-Painting-High-ISO-Review-Camera-Comparison

The first row is SOOC. I found that when developed in LR, none of the three cameras perfectly handled bringing back the highlights from the overexposed light-painted areas. Nikon led the pack with the widest apparent dynamic range.

Check out the RAW files from this test here:

Light-Painting-RAWS.zip (178.3 MB)

Long Exposure Noise Reduction Performance

In my workshops, I am frequently asked if I use in-camera Long Exposure Noise Reduction. To test its strengths, I collected identical exposures of the Milky Way core with all three cameras. ISOs varied in full stops from 3200 to 25600.

long-exposure-noise-reduction-menu-locations

I ran the test first with LENR set to OFF and then set to ON. If you’re not familiar with what this function does, the camera collects the intended exposure that contains both the signal (photo) and the noise from the sensor. It then collects another subsequent, identical exposure with the shutter closed (no signal/photo, just noise) and then subtracts the two. This is also referred to as Dark Frame Subtraction.

Donner-Peak-Milky-Way-Long-Exposure-Noise-Reduction-Tests

Throwing it off slightly was a car that drove up the pass during the ISO 25600 test.

Check out the RAW files from this test here:

Donner-Peak-LENR-Off-RAWS.zip (746.5 MB)

Donner-Peak-Milky-Way-Long-Exposure-Noise-Reduction-Tests-LENR-ON

Check out the RAW files from this test here:

Donner-Peak-LENR-On-RAWS.zip (746.5 MB)

After examining the files from Nikon and Sony, it’s clear LENR does a great job removing noise from these exposures. Canon’s results were decent but not as pronounced as the other two.

Nikon’s LENR performed the best – especially when comparing the tones in Lightroom. The LENR ON file offers a more realistic green cast to the shadows, whereas LENR OFF has a great deal more shadow noise. Furthermore, that noise is magenta in hue.

Long-Exposure-Roise-Reduction-On-vs-Off-Tone

You may find different results based on how you edit the files above to your personal tastes.

The verdict: set this feature to ON if you want the lowest levels of noise in your RAW files with each of these cameras. Note that your exposure time will double as your camera takes the two frames and subtracts the noise from your image.

Four Minute Exposure Star Trails

Sensors heat up over long enough exposures and I wanted to test each camera’s ability to handle this at high ISOs. All three cameras performed remarkably well in this test. They produced images with acceptable noise levels. LENR was set to ON, which helped.

Very-Long-Exposure-Sensor-Heated-ISO-Performance

The Sony had the most random color noise. I have circled it below. This was developed (pushed) in Lightroom. Recall that this camera does not allow High ISO Noise Reduction when collecting RAW images.

Very-Long-Exposure-Sensor-Heated-ISO-Performance-Sony-a7RIII

Check out the RAW files from this test here:

4-Minute-Exposure-RAWS.zip (157 MB)

High ISO Testing of Milky Way Core

After a windstorm subsided in Lake Tahoe, I did some high-ISO tests in the Martis Valley Wildlife Viewing Area.

lake-tahoe-high-iso-exposure-tests-canon-vs-sony-vs-nikon

The results from this test show that the Canon 5D Mark IV lags behind the Sony a7R III and Nikon D850 in terms of image quality and noise levels. Not unexpected, given the 5D Mark IV’s slightly older hardware and technology.

milky-way-core-long-exposure-tests-canon-vs-nikon-vs-sony

lake-tahoe-long-exposure-tests-canon-vs-nikon-vs-sony

The results remained the same when the files were developed. Acceptable images were produced by each camera, with the Nikon and the Sony being slightly better. Below is a comparison of the images, cropped-in and toned aggressively in Lightroom. Note that I only applied a graduated filter to the sky and didn’t use the adjustment brush to follow the contour of the mountains. If one wanted develop an image like this with more accuracy, consider either using the adjustment brush or moving to Photoshop and using Adjustment Layers and Layer Masks. Learn more about how to do this in this video tutorial on exposure blending using Photoshop.

lightroom-edited-long-exposure-tests-canon-vs-nikon-vs-sony

Check out the RAW files from this test here:

Milky-Way-Core-RAWS.zip (667.8 MB)

Very Long Exposures

To test the performance of these cameras with very long exposures, I did a set of 18 minute shots at a low ISO. This is a common technique for night shooters who wish to collect a naturally-lit, very long exposure foreground to lay under a shorter (30 second) exposure of the Milky Way.

very-long-exposure-tests-canon-vs-nikon-vs-sony

Clouds blew overhead during the test, so the stars are slightly obscured. There is almost no visible noise in any of the three files, even at the far corners of the frames in the most shadowy areas. This is likely due to the low ISO and the excellent performance of LENR.

Check out the RAW files from this test here:

Very-Long-Exposure-RAWS.zip (157.5 MB)

Overall Conclusions

Each camera performed admirably at night. They produced images that are more than adequate, especially on small screens. As expected, grain is apparent at higher ISOs. Noise levels were fairly consistent across all platforms, with Canon performing the “worst” (worst being still pretty good). Nikon and Sony performed equally well even at ISO 25600. Both luminance and color noise can be controlled in Lightroom with the latest Camera RAW (especially for those showing their images on smaller screens). For those making massive enlargements of high-ISO night shots, the Nikon D850 or the Sony a7R III are better choices over the Canon 5D Mark IV.

There are slight color shifts inherent to each camera’s sensor – even in RAW, in the Adobe RGB color space, using Auto WB (then equalizing the WB settings in Lightroom). The Sony’s files are noticeably cooler, with more vibrant blues. Canon has the noisiest shadow areas, with borderline-acceptable magenta tint to areas visible when files are pushed hard in Lightroom.

The images from the Sony a7R III have a slightly “digital” look. This is probably as a result of that camera having the highest megapixel count (over 42MP), which gives it the sharpest images of the three cameras. We also cannot discount the importance of the lenses in these tests. The Sony’s GM 16-35mm is by far the newest of the bunch.

I definitely recommend setting Long Exposure Noise Reduction to ON when shooting single-frame still images at night. But if you are shooting a lot of frames, it will quickly drain your battery.

If forced to choose an overall winner, given the above conclusions, to my qualitative eye, the Nikon produced the “best” RAW images at the highest ISOs of 12800 and 25600 when considering color, noise, sharpness, and realism. Both the D850 and the a7R III were able to resolve the glow from molten lava flowing on the Big Island over 100 miles away at ISO 25600. However, when pushed in LR, the Sony image may even be slightly better. See for yourself:

Edited-ISO-25600-Side-by-Side-Comparison-Sony-a7RIII-Nikon-D850

Those are edited. To play with some unedited files at various ISOs from this scene for all three cameras, check out the RAW files here:

Ocean-Scene-Conclusion-RAWS.zip (852.9 MB)

An Important Reminder About Quality Tests

A favorite analogy of mine when I teach photography is that of a chef and an oven in a fine restaurant being akin to a photographer and their camera. Just as the potential quality of a dinner shouldn’t be judged by the make or model of a restaurant’s oven (though a good oven certainly won’t hurt), one should never prejudge a camera as better than another without considering the skills and knowledge of the artist. Keep this in mind when deciding on what system is right for you!

We’ll soon be publishing a very similar post to this one but with an emphasis on video rather than stills. Stay tuned for that!

The following two tabs change content below.
Born in Hawaii, educated in New Zealand, and now living in Lake Tahoe, Grant Kaye specializes in landscape, night-sky photography, motion-controlled time-lapse, and creative filmmaking. His clients have included Red Bull, MSNBC, Yahoo, and many others. See more of his work on his website or join him for a workshop.

Submit a Comment

Your email address will not be published. Required fields are marked *

%d bloggers like this: