scanning film
Flatbed vs Camera Scanning
In my pursuit of film scanning perfection, I initially embarked on my journey with the Epson V600 flatbed scanner. However, after extensive experimentation and meticulous comparisons, I have transitioned to Camera Scanning as the ultimate solution, surpassing the capabilities of dedicated film scanners. This blog aims to objectively present the findings derived from several tests and comparisons, demonstrating the undeniable superiority of Camera Scanning when paired with a 95CRI light source made by CineStill and the Valoi 35mm film holder.
Updates to the article
29 June, 2023: The issue of reduced corner sharpness has now been rectified. I have switched to a different lens that doesn't have such a short flange distance like the native E mount lenses. Therefore, the extension tubes do not severely compromise the corner sharpness. This still is not a perfect solution, and a dedicated macro lens would provide sharper results. However, for the time being, I am satisfied with this setup. I will update this article when I get my hands on a dedicated macro lens. When used with a 26mm extension tube, this lens is able to fill the frame and produce a much higher resolution image. However, the corner sharpness is slightly reduced compared to the 16mm extension tube. It's still better than the previous Sony lens I was using. I have also calibrated this camera and lens combination using my ColorCheckr charts to prevent any undesired color characteristics from either the lens or the camera's color science from altering the final image.
Image corner with previous lens | Sony E 18-135
Image corner with current lens | 26mm extension | Olympus OM Zuiko 50mm
Image corner with current lens | 16mm extension | Olympus OM Zuiko 50mm
Tests and Comparisons
Epson V600 in top row, camera scans in bottom row. CineStill 800T. The link to all the full res images are at the bottom of the page.
Let's proceed with some one-to-one comparisons. The test shots below are all from the same CineStill 800T strip. I captured the test chart in daylight, despite the film being tungsten balanced, as I was working on a PowerGrade for DaVinci Resolve to recreate that specific look. For this test, I repurposed these negatives. I used Negative Lab Pro (NLP) to convert all the images and made no adjustments to the NLP output whatsoever. The rationale behind this approach was that colors are subjective, and either file can be edited to resemble the other in terms of colors. However, if one option consistently delivers subjectively better colors right from the start, that should be the preferred method.
CineStill 800T overexposed by 2 stops | V600 scan
CineStill 800T overexposed by 2 stops | Cam scan
CineStill 800T overexposed by 1 stop | V600 scan
CineStill 800T overexposed by 1 stop | cam scan
CineStill 800T exposed neutrally | V600 scan
CineStill 800T exposed neutrally | cam scan
CineStill 800T underexposed by 1 stop | V600 scan
CineStill 800T underexposed by 1 stop | cam scan
CineStill 800T underexposed by 2 stops | V600 scan
CineStill 800T underexposed by 2 stops | cam scan
Based on these samples alone, it can be definitively concluded that the camera scanning method surpasses the Epson V600 flatbed scanner in terms of latitude and sharpness. The camera scans produce significantly sharper results, offer superior latitude and flexibility, and exhibit no noticeable digital noise. Conversely, the flatbed scans appear soft to the extent that they almost appear out of focus, likely due to the flatbed's inability to focus. Additionally, the flatbed scans suffer from significant digital noise, including color noise, and demonstrate a noticeably inferior dynamic range. It is important to note that higher-end scanners such as the Epson V850 or V800 do offer finicky methods to move the film plane into focus, but these scanners are considerably more expensive and, as I will discuss later, still fall short of camera scanning solutions.
Furthermore, in my personal opinion, the flatbed's color reproduction also leaves much to be desired. This can be illustrated through the following examples. Keep in mind that though I personally prefer the colors I get from the camera scans, colors are of course very subjective and we may not reach an agreement regarding that.
CineStill 800T | V600 scan
CineStill 800T | cam scan
The limited dynamic range and presence of digital noise are clearly apparent in these images. However, it is important to acknowledge that these issues are not the fault of the scanners or Negative Lab Pro. The images were captured under cloudy conditions, and in addition, the film used was tungsten balanced. This specific image is part of the same film roll as the previous test shots.
In the V600 scan, the clouds in the background lack any discernible detail, and there is a noticeable harshness in the highlight rolloff. Similarly, the darker areas also lack detail and are noisy. It should be noted that this noise is digital noise, distinct from film grain. Additionally, the V600 scans exhibit prominent digital color noise and peculiar artifacts. The discrepancy in detail can be further exemplified by the following example.
Ilford HP5 400 | V600 scan on the left and Camera scan on the right
In this image, it is also very clear that the camera scan produces tack-sharp images and, depending on the camera used, much higher resolution as well. Currently, I am using a Sony a6600 camera with a 24-megapixel sensor. However, at the moment, I don't own a macro lens. My current solution is not ideal - I'm using a 16mm extension tube - and it results in reduced corner sharpness. That said, this issue isn't major, and even with the reduced sharpness in the corners, the images produced are still far superior compared to any flatbed scanner. Upgrading the camera and lens will surely surpass all but drum scanners.
Speaking of which, Thanks to a fellow photographer Ankit Dey, I got my hands on his Nikon D850 and a dedicated macro lens. The D850 has an excellent 45 megapixel sensor which, in conjunction with the macro lens, can produce 37 to 40 megapixel scans even for 35mm film. Thanks to the dedicated macro lens, The images are tack sharp corner to corner. Here are some images scanned using that camera and lens setup. The light source and film holders are the same as the rest of the images in this article. The link to the full resolution files are, again, at the bottom of the page.
Nikon D850 scan (L) & 100% magnification in 2000x2000 frame (R) | Tokina 100mm f/2.8 | image by Ankit Dey
Nikon D850 scan | Tokina 100mm f/2.8 | image by Ankit Dey
100% magnification in 2000x2000 frame | Tokina 100mm f/2.8 | image by Ankit Dey
Now, let's compare the colors you get from either system. The following examples, like all the images on this page, are straight Negative Lab Pro outputs. I only white balanced the raw files off the film border for the camera scans and cropped the images as necessary. Other than that, I have also applied the best sharpening settings that I have come up with over the years of testing and experience.
Kodak Colorplus 200 | V600 scan
Kodak Colorplus 200 | cam scan
Kodak Colorplus 200 | V600 scan
Kodak Colorplus 200 | cam scan
Kodak Colorplus 200 | V600 scan
Kodak Colorplus 200 | cam scan
For my current setup, I have calibrated my camera for my film scanning environment using my ColorCheckr chart and its companion software. While Datacolor is notorious for being, to put it mildly, not very accurate, I think it still makes the Sony colors match the real world much better than stock. The goal of this step is to ensure that the camera is capturing and reproducing the colors in the film negative (or positive for that matter) as accurately as possible so that NLP has a more accurate input to work with. Here are some examples of images with and without those calibration profiles. Note that the images that do not have the word "calibration" in the description do not include those corrections. The reasoning behind not applying the correction profiles was that the majority of people will not be calibrating their cameras for their scanning environment, nor is it that necessary for the vast majority.
Kodak Vision 3 250D | cam scan | without calibration
Kodak Vision 3 250D | cam scan | with calibration
Kodak Vision 3 250D | cam scan | without calibration
Kodak Vision 3 250D | cam scan | with calibration
With that, I think it's now safe to say that scanning with a camera produces better colors and tones than an Epson V600. But what about a more expensive flatbed? Let's now look at scans from an Epson V850.
Kodak Colorplus | V850 scan
Kodak Colorplus | cam scan
Kodak Colorplus | V850 scan
Kodak Colorplus | cam scan
Kodak Colorplus | V850 scan
Kodak Colorplus | cam scan
Alright, then. What do you think? Personally, I prefer the colors from the camera scans. However, the differences, at least in these images, are minor and can easily be edited to resemble either. That said, the camera scans still offer more detail and have no trace of digital noise, whereas even high-end flatbeds will have slightly lower resolving power and much worse dynamic range. While this lack of dynamic range may not matter much for scanning negatives, as film negatives don't typically have a high dynamic range, I find the camera scans to be more flexible and easier to work with. The file sizes with modern camera raw are also more manageable compared to high-bit-depth TIFFs. Additionally, I'm not a big fan of VueScan or SilverFast DNG outputs as I feel they don't provide much more flexibility compared to high-bit-depth TIFFs.
Moreover, with modern cameras that support pixel shift technology, it's possible to achieve pixel-perfect colors with abundant detail and very high dynamic range. However, these factors only come into play when creating very large, high-resolution prints. For the vast majority of people, even a lower-end modern DSLR or mirrorless camera will produce brilliant results that surpass the majority of dedicated scanners. Even when compared to Noritsu or Frontier scanners, I believe the camera scanning method is superior because newer CMOS sensors offer incredible dynamic range, latitude, and a very low noise floor. The ability to fairly easily calibrate the capture device to accurately represent the scene colors is also a substantial advantage for me.
So, this is why I made the switch to camera scanning and sold my trusty Epson V600 this year (2023). Have I convinced you to switch as well?
Film Resources
