Under exposing your photos do not make them better

SUBSCRIBE to this KZbin Channel at www.youtube.com/@edinburghpho...
SIGN UP for my newsletter at bit.ly/EPW_SignUp
VISIT edinburghphotographyworkshop.com
----------------------------------------------------------------------------------------------------------------------
I recommend Kase Filters - find out more at bit.ly/EPW_Kase

[@FreshLoL]

I believe underexposing in some situations can be a good thing. If you enjoy photographing during the night, it can lead to preserving the highlights much better, and if you are not concerned by having too much detail in the shadows then it works just fine.

[@_SYDNA_]

Maybe he's right. I generally try to leave a little more headroom on the top end though to avoid accidently blowwing out the upper end.

[@simonmuller3706]

I absolutely agree with most of your suggestions. However, in the grassy hill photo I actually like the silhouette effect of the original edit. With a few selective tweaks to the sky that photo could almost look like a painting. I think it is important to mention that aiming for a perfectly balanced histogram is not always the way to go. Clipping the shadows can be an artistic decision if done intentionally.

[@JerGoes]

I agree with much of what you're saying, however artistic vision and intent Francis had may have been to have the hillside and figure be a silhouette and not exposed, personally I like the touch-up of the sky but prefer the dark silhouette of the hill and figure. My DSLR doesn't have a onscreen histogram (until the photo is taken) so I'm working with the light meter and I'm forced to use my own judgement based on the colour of the subject vs the background as well as experience using my gear to gauge what I'm likely to get. Even during post and editing I rarely look at the histogram, sometimes I want blacked out or blownout areas I'm more concerned about what I want the image to look like, you are correct deliberately underexposing as a matter of course isn't good advice.

[@WillNewcomb]

Interesting. I'll have to do some trials myself to see if I can noticed the difference, but I only underexposed by 0.7 of a stop. I don't often use the histograms but perhaps I should as it'll give me more info about my images. I use an iphone12pm. Thanks

[@cantkeepitin]

Good video, but add more examples with highlights like at night in the city

[@paulhenry7]

Underexposing your photos does not make them better?

[@richardchapman8855]

The title do not make a good sentence

[@henrikmartensson2044]

Some good advice, but it does not apply everywhere. The problem is that you are providing context specific information without providing the context. Here are some contexts in which your advice does not apply: * When you want to give colors a bit of extra saturation in camera (Useful in nature photography) * When you want to protect highlights (for example you have large dark areas in the frame, that trick the camera into overexposing, or you just have a lot of contrast) * Low key photography * High key photography * HDR photography * The Film Noir genre * The Neo Noir genre What you are saying is perfectly fine, in some contexts, but it is not generic advice that fits in any context.

[@L.Lyubomirov]

I have to say NO. Underexposing a photo can be done with a camera with good dynamic range and ISO invariance. For example... I use a Pentax K1, if I under expose at -4 stops I don't have any problems because my sensor is ISO invariant and it has a great dynamic range, I also have a very good chance of shooting handheld most of the time! So underexposure is great if you have a great sensor and know what your camera can do,you must know your limit.

[@_SYDNA_]

And, while under-exposure may not make a net difference (or may), over-exposure can make an image irretrievable.

[@L.Lyubomirov]

@@_SYDNA_ Yes this is true with digital overexposure is not that good as with film era.

[@edinburghaction5515]

The histogram in-camera shows an 8 bit jpeg so it's not an accurate representation of the data captured by a camera sensor.

[@edinburghphotographyworkshop]

That is correct, although with a little experience of reading the histogram on the back of the camera and comparing it to the Lightroom histogram, you can get a good indication if the image is being overly clipped. There will always be more data captured than is seen on the back of the camera histogram.

[@antonoat]

Under exposing your photos does not make them better !

[@dance2jam]

First time to the channel, Mr. Dyson. Thanks for your outlook and explanations. I'm an amateur photographer that has not been shooting a very long time. That said, I do not agree completely with your explanation, and I'm happy to explain why. I'd also love to hear your thoughts on my comment - as that is how I learn - and I don't mind a good conversation around photography. 1st: I understand where you are coming from with ETTR. If I get a bit wordy here it's not for you, but those reading the comment and wondering what we are talking about. Exposing to the right (ETTR) provides more photons (data) to create the image. The first assumption I think you make is that we then want to edit the image with post processing - using the additional information in the shadows, midtones, and highlights to tweak it to our satisfaction. Understood. There are two reasons your argument isn't exactly sound. A. There are the cases when you get it right in camera and your shooting to crush the blacks (i.e. ETTL) or protect the highlights because of very bright subject or lighting. Blown highlights is lost data, compared to a lower signal to noise ratio (but retained data) in the shadows. Yes, I realize if you ETTR the image and then bring the highlights down (assumed not blown) you have more data in the shadows - but if you don't want to process the image much (for what ever reason) and you capture the look you are going for, then underexposing may be appropriate for "some" images -depending on taste. My second issue is with how you personally arrive at your conclusion in this video: And please hear me out on this one. My first assumption is that we are talking Digital Photography. Early in the video (prior to the 3 3/4 minute mark) you lump a number of issues together. A. Underexposure vs ETTR, B. The effects of increasing exposure. And C. The amount of data underexposure vs ETTR present in the file size. I get it, but the way you present the argument is flawed in my opinion, and here is why. 1. You talk about increasing exposure 2 stops by decreasing your shutter speed 2/3rds of a stop (fair enough) and increasing ISO 1 1/3 stops. This last statement in my view, is wrong on a number of fronts. A. ISO is not part of exposure in the majority of cases. It is gain and amplification of the Signal to Noise (S/N) ratio after exposure has taken place. Exposure is the amount of light/photons captured by the sensor because of the size of the aperture or speed of the shutter or sensor capture. That's it. To maintain the same exposure, you balance that seesaw. If you want better exposure, more photons/information in the image without clipping the highlights, you need to slow your shutter down (which you did), or open your aperture (which you did not). Increasing ISO just produces a brighter image, but does not contribute to exposure directly. On that note, if your goal was ETTR, and all you did was increase your ISO by 2 stops - you've done nothing to improve the image or improve exposure prior to post. You've just increased the brightness of the image - and in an ISO invariant sensor - it wouldn't matter if you shoot underexposed with ISO 100 and bring it up to ISO 400, or shot at ISO 400 to begin with. Underexposing and raising the ISO 2 stops gives you the same S/N ratio as if you shot at ISO 400 to begin with. In your example, there would be "more data" because you slowed your shutter speed down - and increased exposure - i.e. photons/signal. Had you left ISO alone, and opened your aperture up 1 1/3 stops, you'd have an even greater exposure (or information) without magnifying the amount of shot noise in the shadows. None of this may be relevant to the finished look of the photo, but it could in many circumstances. In the end, my view is that if you want to increase the amount of data/photons/information in your photo, then you need to increase your exposure by either opening up the aperture, and/or slow the shutter speed down. That will increase the signal to noise (photons to shot noise) in the shadows - where information is thin, and in the highlights where information is plentiful and everywhere in between. ISO plays no role in this or ETTR (EXPOSE - to the right). It just brightens your image and raises noise with it. I bring this up because I believe the intent of your tutorial was to increase information in the photo, and raising ISO as part of your argument just complicates the picture (so to speak). Your thoughts?

[@edinburghphotographyworkshop]

You would be correct in what you were saying if noise was added at a linear rate across the image. That isn't the case. Because there is more data in brighter areas, the effect of noise is more obvious at the darker end of the histogram. In one of my camera club talks, I demonstrate this with two images taken in a dark mud hut in Africa. Using ETTR and increasing the exposure by adding an extra two stops of ISO leads to an image that the majority of viewers have agreed appears to have less visible noise.

[@dance2jam]

@@edinburghphotographyworkshop Thank you so much for taking the question and responding. I have many questions about the parameters of the shots you talk about, but suffice it to say that when I dig out from hurricane Beryl aftermath, I'll experiment with this concept. Just curious about the science here. There are many opinions on this (including David Bergman - Canon Explorer of Light) who recently stated in his opinions that he felt ETTR was over-rated. I'll try this under several circumstances. Thanks for the response.

[@user-eh8jv2em2o]

​@@dance2jam couple of things worth discussing: linearity of RAW data, bit-depth/precision and ISO's analog gain. RAW data is linear. That means any exposure offsets to it do not change the inner contrast of the data, do not compress highlights or anything like that. With 14 bit RAW each pixel has value of 0 to 16000 (roughly). Why would anyone want his data to occupy only lower part of this range (like 0-5000) while inevitably clipping more blacks and leaving empty space above? Why you can't just fix this later: because these values are discrete, and for -1 EV example: take [1,2,3,4,5,6,7,8] divide by two, chop off the fractional part then multiply by 2 again. You'll get [0,2,2,4,4,6,6,8]. That's your banding, lost precision, you can't restore original values with "exposure" slider. So ETTR isn't some trick or religion, it's just the only mathematically sensible approach. It might not be the "safest" approach (because your whites approach clipping zone) but you can set up your clipping indicators, watch the histogram, in other words it's possible to learn what's safe for your camera. And let's not forget that even if whites were slightly clipped they usually clip in one of the channels, so restoration is often possible especially when clipping area is small and doesn't have much variety in colour (typical for skies). Next, ISO in most cameras is not purely mathematical (or digital). There's analog gain and digital gain. Analog gain means: more voltage is sent to the ADC. Digital gain is just math on chip, it adds no noise (but adds no real precision either). Typically between ISO 100 (sometimes 200) and ISO 1600 there's only analog gain. Now, why it's important: ADC has certain noise floor, voltage threshold. This is separate from sensor's noise: sensor has noise but ADC has noise too. So more analog amplification to signal means that ADC noise isn't as significant in comparison. That's why 1/60, ISO 200 (underexposed) pic will have more noise than 1/60, ISO 400 pic. Because of signal to ADC-noise ratio. And the 1/60, ISO 400 pic will additionally have better bit-depth (explained in 1st part of my comment). To put simply: result = round_to_ones(analog_gain x (arrived_light + photon_noise + dark_current + thermal_noise) + adc_noise) x digital_gain. Photon noise means randomness of photon arrival, typical for weak light. This is overly simplified since in reality with larger megapixel count adc_noise is less relevant after image is downscaled to final size, also pixel (sensor) area matters. In editor consider that you're changing digital_gain only.

[@dance2jam]

@@edinburghphotographyworkshop One of my questions that I wanted to ask you: Are you shooting on the Canon R5, or more directly, and ISO variant camera?

[@edinburghphotographyworkshop]

@@dance2jam My main camera is the R5, and I also use an R6 Mark I as a second body

[@gigafish2x077]

To the left, to the right, it's not politics, you should learn the exposure triangle and expose to get all the information for it. 67-90% of the time, it will be to the left, the other to the right, with the rest of the exposures at 0 ev.

[@IanKnight40]

In the example of the lilly you have increased the iso to 125 . Is this not underexposing the image because of this?.

[@edinburghphotographyworkshop]

Increasing the ISO value will add light onto the sensor, so no.

[@maggnet4829]

​@edinburghphotographyworkshop Nope, increasing the iso will not increase light on the sensor, but rather the opposite. It will artificially amplify the signal, which causes faster shutter speeds and thus less actual light falling on the sensor.

[@edinburghphotographyworkshop]

@@maggnet4829 I think you are misunderstanding that while I am increasing the ISO which will increase the EV of the image, I am not changing the shutter speed. As a result, the image will appear brighter, and the areas which contain the least data (i.e. at the dark end of the histogram) will have shifted to the right where there is more data. This results in less visible noise despite a higher ISO value.

[@maggnet4829]

@edinburghphotographyworkshop This only applies to compressed formats like jpg. raw files store the data in a linear fashion, so no matter where you push it, it's the same. Some sensor however have a dual iso. In some cases, it can be beneficial to go for a higher iso in that case.

[@mattstich7979]

underexposing does not cause an increase in noise. it's only if someone is trying to bring back shadows. I also would say the histogram is a tool and not to bog down your art by just relying on that. It's often easier to bring back a darker image than an overexposed one. Most modern cameras are a bit bright and underexposing helps with this. Depending on the situation I often underexpose by 0.3 to 0.7 You also can't rely on file size as "more or less" information. As an image with gain will give you much more data and that doesn't mean better. So that is hardly "bad advice"

[@edinburghphotographyworkshop]

Respectfully, I am going to disagree. In one of my camera club talks, I have shown that by shooting a scene in a dark environment and using the expose to the right technique with a higher ISO value, there is less noise than shooting the same scene under-exposed with a lower ISO value. As I show in the video, there is less data at the darker end of the histogram, which is why we see more noise. If you move the exposure toward the right, the lighter (not over-exposed) areas contain more data and therefore less noise.

[@BPetiBP]

lol ...😀