Don't Use "Flat" Picture Profiles!!!
Вставка
- Опубліковано 24 лют 2017
- Log and Cinelike picture profiles are perfect solutions to getting more dynamic range from your camera, right? Wrong. In this video I'll discuss the problems with flat picture profiles and how to avoid them.
►Follow Tall Guy Films
Facebook: tallguyfilms
Twitter: tallguyfilms
Wordpress: tallguyfilms.wordpress.com - Фільми й анімація
For those of you who still don't believe my video, here's an explanation from a professional cinematographer.
www.redsharknews.com/production/item/4841-this-is-why-you-shouldn-t-use-log-gamma-all-the-time
Tall Guy Films Good video, and thanks for sharing this link-very informative
Did you actually READ the article you linked here? The author specifically states he uses LOG on the Sony a7 series of Mirrorless cameras when the situation calls for it. He does NOT IN ANY WAY suggest you shouldn't use log on them... as you clearly do.
@@5MinuteGuide That's not what he said in the video. Did you happen to watch the video? Or just respond to the title?
@@5MinuteGuide I suggest that NO ONE should ever shoot log or flat on any camera that outputs in a color bit depth lower than 10 bits. The basic math dictates this, not some sort of ideological bent. When you have 8 bpc (256 levels per channel) you have to be a fool to push that data only to pull it right back! Changing gamma curves is a lossy operation, and log gamma is a pretty extreme curve change (visible as the very poor contrast). www.jodybruchon.com/2019/02/20/no-this-doesnt-look-filmic-shooting-log-flat-and-luts-all-suck/ and if you don't want to believe me, you can read www.imaging-resource.com/PRODS/sony-a7-iii/sony-a7-iiiVIDEO.HTM which says: "The A7 III at best shoots 4:2:0 in 8-bit internally, and 4:2:2 in 8-bit to an external recorder via clean HDMI out. That's less than ideal, and even if you're oversampling or doing all that jazz with the sensor, that kind of color compression means you're not going to be able to take full advantage of those specs. Most colorists will tell you that shooting in Log isn't even worth it if you're not at least in 10-bit."
@@JodyBruchon thanks. this makes sense. what settings do your recommend in natural profile for the G85 (saturation, contrast, noise, sharpening, shadows, highlights) for the greatest dynamic range?
The title should “becareful when using flat profile” NOT “DONT USE FLAT PROFILE”
The science behind this is spot on and I applaud you for addressing it. For me personally, I end up doing some pretty extremely stylized grades in post so I'll continue to shoot flat for that flexibility later but the information you've gone over will be kept in mind when shooting 8 bit.
thank you for making this video because I always felt like I was losing information and creating more work for myself by shooting in a flat profile. This clears up a lot for me. I used to shoot standard often but then I found out about flat profiles and shot with that all the time and under every situation and I never felt like it enhanced any of my images just creating more work for me to get the same look I would have gotten if I just shot in standard.
Wonderful explanation thank you! I've been mimicking a flatter profile on my sony a6000 which doesn't have any log option by making my own profile adjustments and I have often wondered if I'm just making things worse for myself later. There are some cases where it has helped me out(I think) and some cases where it made the noise nearly unbearable. Luckily I have an external monitor on the way with some awesome assist features to help me get better exposure in the first place. Can't wait to see how that helps and put this new information to use.
I totally agree
I have been in electronic design for many years particularly in digital systems
Putting your point another way
we have 256 values in an 8 bit system
with REC 709 the HD standard we have a maximum of 6 stops
so each stop has approximately 42 values per stop ( In practise they are not allocated equally mid range gets more values)
If we have 12 stops the values per stop drops to approx 21
this has a serious effect on the fidelity of the picture
In engineering we know if you gain on one thing (ie greater dynamic range) then you lose on another --
and yes LOG needs at least 10 bits
Great video, and great information. I'm familiar with all of that for years but I'm always surprised how many people don't realize that going too gray is just tossing away good information.
Tall Guy Films I am using Sony a7s ii for wedding, while shooting in slog2 or 3 picture is too grainy and while editing in adobe premiere it's too difficult to do color correction.
No idea why this has so many downvotes. I've been working professionally in video production for 10 years and everything he says in this video is accurate. You could argue the title is misleading but I have no idea why you're all so upset about this video honestly.
Probably because of the title and the baity thumbnail. Very reasonable points on the video itself though.
you are right about that flat profile. I do occasionally get difficult situation when remastering my flat video during post production and just couldn't get that good quality. The contrast, brightness, saturation just wouldn't work and the scene is just look bad. The outcome is that when I apply LUT, it won't look good. Thanks for that great advice. You are right, do not force yourself to use flat profiles for everything. Sometimes using vivid or nature profiles saves you lots of post production.
Brilliant video. At the end of the day it is a matter of taste. I have tried many times shooting on Cinelike D - 5 with every setting (contrast, sharpness, saturation and noise reduction) on my Lumix G7 and on the same day, same light situations, some scenes look better than others. Some videos from Cinelike D cannot be used at all, even though I got the best exposure using the histogram and a good lens. But now I am only filming using everything -5 on Natural profile. It is way easier to work with and it seems every picture looks fine and easy to colour grade / colour correct.
I wish I watched this before I destroyed my film with Flat picture style 😫
its only destroyed if you haven't learned how to color grade. Using flat profiles doesn't mean you have a terrible picture filled with noise. not at all.
I have been shooting all these years in flat profile because every youtuber to me to do so (giving me reasons why I should). It added nothing much to tbe value besides adding a tonne of extra work to my workflow. I don't do fancy colour grading. I started using portrait profile recently and it's such a relief for me. My clients wanted sharp, vivid colors and they don't really care about technicalities. Whatever you have mentioned here, I have been wondering myself from last few weeks myself. Now I spent time shooting more and editing less. Time is money
It's funny because his video looks like it was shot log. But I do agree here. The standard look (rec709) looks fantastic and will work in most all situations unless you are dealing with a very naturally contrasty environment. If you are controlling your lighting, I definitely agree to not shoot log. Some situations where log makes total sense is night time in the city, infront of windows. I think for professionals who know exactly what they are doing with log, they can handle it in any situation. But for beginners (talking no experience, all the way up to 5-10 years of experience) Matching log footage for a coherent colour edit, is alot harder with log. Not to mention getting wildly different exposures during the shoot.
CalviNNation It's not even about professional knowledge, it's an issue of cameras not storing enough color data to make log shooting without footage damage possible. The GH5 is the first affordable DSLR with 10-bit color output in the camera. The vast majority of DSLR shooters have 8-bit output files on their cards and that format doesn't provide the required latitude to use log and flat profiles without serious issues. Anyone with a camera under $2000 is guaranteed to be unable to use log/flat profiles without noticeable loss of quality. Graded flat footage looks unnaturally over-saturated because that's the only way to hide the damage: make it look like a stylistic choice instead of a fundamental problem with lost color data.
Yep, shot flat or just graded, so that his acne isn't as visible. You can tell by the skin tone.
Night time? What about noise?
Couple of things:
1. Noise isn't "created" by bringing up the shadows, nor is it reduced by exposing to the right. It's simply less prevelant, yet it is still there. The amount of noise generated at iso 100 will be the same whether you over- or underexpose (and it will be throughout the image, highlights and all).
2. An 8-bit camera holds 8 bits of information per color channel (red, green and blue respectively), thus a 10-bit camera will hold 64 times the color information of an 8-bit camera (not 4!). This is a nitpick as it just reinforces your point.
Yet, (and I'm not sure how this pertains to video), in still photography changes applied by the color profile will be written to a side-file and not affect the raw data of the image. This is a non-destructive way of working, ensuring - for example - that your flat color profile doesn't actually discard data and nuances in color values.
No if you use dlog it's gonna give you a lot of of noise in the dark areas
Your 8-bit/10-bit explanation could use some refinement, so I'll try to explain. 8 bpc ("8-bit") holds 64x the information (not entirely true unless using full-range YUV or using RGB, but Panasonic at least supports full-range YUV so I'll give you that) but that is a figure across all channels. On a per-channel basis, it's 4x more information. Filming a very red object, for example, will invoke as little as 4x the information (talking RGB for simplicity in understanding).
The information multiplier is not that meaningful in practice, though. 10-bit doesn't record levels that are 4x brighter than 8-bit, it records 4x more fine detail than 8-bit. It's helpful to think of it with respect to the final output file which is almost always 8-bit, specifically by thinking of it as a fractional part with 1/4 granularity, and in post with no color changes at all, this fractional component (say 133.75) is effectively just dropped or rounded, leaving the integer (133 or 134) you would have gotten in 8-bit anyway. Where 10-bit helps out is when several luma/chroma change operations are done on the clip in post. Each layer of effects processing has the potential to introduce rounding errors. When there is more information beyond that of an 8-bit channel, the rounding is more precise. Most editors now process data in 16 or more bpc internally to mitigate this issue (thus why video and image editors got significantly slower in the 2010s), but there is still room for rounding errors because 8-bit data has already been quantized at a lower depth than the internal processing is capable of, and those quantization boundaries become banding when sufficiently "amplified." Flat/log shooting requires such "amplification" of subtle values to restore a normal gamma curve, which is why it always looks like hot garbage when done with 8-bit footage. 10-bit footage has extra data below the 8-bit quantization boundary, so "pulling up" those values doesn't easily result in banding since there is real data there, not a "hard line" under which there is no data and flat zeroes must be assumed. You can still cause banding, but you have to push harder. There is also the question of how well the sensor in use can actually discriminate between the levels of extremely fine color variance involved, which I imagine is only a concern at extremely high ISO settings or with near-blown blacks and whites.
It's also critical to remember that the compression used on most cameras is lossy and is designed to discard subtle detail to reduce storage size. These algorithms use tiny squares of varying sizes to scan for discardable fine changes in detail, and extreme shadows and highlights are particularly favored for this process. One of the things that flat/log tries to do is to push these extremes away from the extremes so the algorithm doesn't consider the changes as discardable, but in doing so, they crush the subtle differences in the midtones into even smaller differences, making subtle midtone gradients (you know, like *skin tones*) appear to be better candidates for discarding details. The result is banding in midtones that is immediately "amplified" by the curves required to pull the values back to normal. This makes the prospect of using such imaging trickery even more foolish on consumer hardware, because such hardware tends to favor smaller file sizes, lower bit rates, and cheaper processing hardware (read: lower efficiency per unit of video bit rate), resulting in more compression macroblock banding waiting to be revealed in post.
Get it right in-camera. That's what everyone should do.
Just some thoughts; I read through the article, and still to my understanding, a DSLR's sensor offers greater bit depth than what it is allowed to record and thus setting it as flat as possible will still offer better detail in the extreem shadows and highlights.
Your point is valid when shooting in a controlled set where dynamic range of the scene itself is narrow and thus shooting flat wont recover additional detail (and yes, probably debilitate micro-contrast). But this point does not mean we shouldn't shoot flat as the title of this video would suggest (and thus was the probable cause of the dislikes). A title like "should you always shoot flat?" might have have a less triggering effect on viewers.
Good luck going forward though. I still appreciate the thought on the subject.
actually, that's a good point - people talk about using flat profiles in order to more "evenly distribute" the range of values but that would probably have the effect of creating some aliasing. it would be useful tho to see what you're talking about with examples tho. most of the log footage I've seen looks pretty good.
You are right. There are too many flat profile tutorials for cinematic look on youtube. Got me confused. Your explanation helps a lot.
Just got thur testing your ideal.. Thanks for making me re think shooting in flat always.
Depends on the camera. High bitrate cameras with good dynamic range can benefit from flat profiles. There is a reason why cameras like RED, Alexa and BMC tend to use very flat profiles.
They're not FLAT... they're using log encoding which captures ~ 14 stops of information, and sprinkles it (depending on the type of log curve will decide how many code values each stop is getting) into a 10bit, 12 bit tonal space.... Your monitor, being only about 6 - 7 stops REC709, can't display this properly, so it appears very flat thus, one must use a display LUT. Almost all delivery is going to be 8-bit Rec709/SRGB anyway in the end unless you're going DCI or something greater. It's not so much 8 bit, it's also the Chroma subsampling. 4:2:0 is terrible... 1 pixel out of four is only accurate. So no matter what profile you're using, 4:2:0 is going to ruin it anyway. This is what causes the blockyness and blotches (and other types of artifacts) you see in 8 bit S-log2/3 profiles. You have to tailor the scene to the camera really... know the limits and make it work to convey what you want your audience to see. If you do a lot of high end vfx, 4:2:0 will rear it's ugly head all the time... sure there are some tricks to mitigate this, but shooting RAW/LOG 12bit it the way to go... Anamorphic ;) Arguments about 8 bit 4:2:0 being fine and displays well on the big screen is just the poor mans illusion to back up his craft... eventually he/she will be using a RED, Arri, Panavision film or whatever high end camera in the end... then when they do that, you'll hear a change in tone. DSLR's have there place, and Shane Hulbert proved that, but he embraced the limits of the 5DM2 to fit the shot he was shooting (without using ML and shooting RAW which I don't think was avail at that time....) I own a RED now and I also shoot on 16mm film... and shooting on film is where you really get a very deep understanding of it all... especially lighting. There's no UNDO button when shooting film... you gotta know what you're doing or it will be very expensive.
It has to do with how much dynamic range you can capture and how much control you get. If i don't want a contrasty image or I find the artificial sharpness that a standard Profile offers bad, then why would I not experiment with my profile. I know how RAW and 10 bit works, I've been using a BMPCC for the past 4 years. But what I've learned is that you have to get the best out of what you have. I've moved to a Panasonic g85 because I know that if I know what I am doing, then I'll be able to get a good image. Downscaling the footage from 4k to 1080p also gives you more wiggleroom. But saying that you NEED a RED, Arri or other of the big boys to get a great image is an insult to yourself. Your camera isn't the one creating the movie, you are. If you're horrible at lighting and doesn't know how to use a camera, then any camera can produce a garbage image. Sure I would like a 70000$ camera, but before I get into the industry that simply isn't realistic. GH5 is a good step in the right direction with 10 bit 4k recording, but it isn't a must if you know what you're doing with your image.
Nikolaj Christensen I just bought a G85 a few days ago. If you could spare a few moments, what is your set-up to get the best image out of it?
This is really an example of being different and swimming against the current, but totally same. He's just saying that shooting log is great on best high-end cameras but faking it (using cinestyle and the likes) with low-end camera will end up trash due to low dynamic range. :)
Those of you who are confused to shoot for a project, I am doing a lot of research and for some reason felt I should shoot in log to be more professional. But then I came across a fact that if you just shoot out of your In-camera Creative style to Standard or Neutral , then just reduce your Contrast, Saturation and Sharpness to (-3). It helps make your colours go flat and helps you do basic Colour Grading and get the cinematic look when you apply any luts.
Also nice video bro , haters are just going to hate and make fun so fuck em. Keep going hard !!
Despite the downvotes from people who got their filmmaking training from a bunch of fake "pros" that really just want to sell them LUTs, I've parroted this for years and I fully stand behind you on this issue. I have written numerous articles on the subject on both my personal blog and my Gazing Cat Productions blog in an attempt to convert people who have been misled by the "shoot flat" false prophets, and sometimes I get a "this was helpful" but it's mostly a thankless job. I salute you for this video.
Please tell me who are the true prophets of videography/photography?
I need to follow them
Im so woke now and I stopped following hipster false prophets of videography/photography
Great video and explanation! Thanks for breaking it down -- it was very helpful. 👍
good stuff man, really well stated and I appreciate the way you laid it all out
all valid points. Just get the BMPCC 4K and shoot raw 😊👍
Thanks for the video. Does anybody know if the various Cine gammas in sony cameras have that same color compression issue as the SLOG gammas???
Tis a great video - I was just thinking this myself.
CINE profiles on a7sII seem like a good compromise for most shooting.
Love how you lit the scene for this video. Can you give us some insight into it!
i just bought a Canon 2000D dslr rebel T7 which picture profile should i use. should i just use standard or neutral or should i adjust the settings myself if thats the case what settings should i have to get the best possible result
I thought I was gonna see some actual examples. :/
Thank you. Do you have any suggestion for sony a6100? It doesnt have any picture profiles...
thanks for the useful info.. it would be nice if u can make more videos around this topic. I have lumix G7 and still not sure what profile should I use for day to day filming? can you share your recommend video profile settings? thanks!
If you check out my channel, I have a video on what settings I use with the Lumix G7. In general that's the "Natural" profile, but there's some more information in the video.
Thanks for this great explanation!
Many people shoot flat on the GH5 with
most settings at -5.
How do you sharpen fast in post?
Best regards
Shoot on Panasonics with everything at 0 except noise reduction which should be at -5. Noise reduction reduces fine detail and NR can be applied in post if you want, but the lack of NR is part of what makes raw video look so much better. I have a video about how noise and fine detail are identical to a computer that demonstrates what NR does to a photo if you're interested in seeing more.
I have a m50, is it better to stick with the neutral profile instead of the cinestyle and eoshd c?
so what would you recommend for me using my a6300 to shoot video on so I don't have to Color grade and still get great skin tones
ok, being a computer nerd just getting into photography and video production (for fun not to be a youtuber), I can see how a cine "type" profile artificially resets the black and the whites to something less. I also understand that lower quality cameras like mine, dont have a "cine" style profile. It doesnt have enough bandwidth to justify using it like the high spec cameras do.
Would lowering the contrast and saturation in the settings "compress" the data like just like a "cine" profile or not? Isnt that asking the camera to do something it is already capable of doing?
Welcome to the photo/video club! I started as a computer nerd myself.
Typically when a camera has contrast, saturation, or sharpening controls, they're a bit tricky. It's just like adding any of those things in post, except they're "baked" into the image, meaning they're added before the image is compressed.
Take contrast, for example. Having more contrast will help show fine details in areas of the image with low contrast, but you'll lose detail in areas with high contrast. Lowering contrast will do the opposite. Since that processing is done with the raw image before it's compressed, you can't really change anything in post. For those types of settings you'll have to test with your camera which options give you the best results. I keep my Lumix G7 on "0" for all of that.
And you're probably seeing at this point why RAW is so great, because it gives you all of these options in post-production.
Yeah I can see your point. I have the lowest of the low when it comes to DSLR cameras...its my training camera. I am talking about a Nikon D3300. The price was right at the time.
I have also seen the videos that instruct someone to flatten the image but when I have tried it, it seemed a bit aggressive to try to bring everything "back" in post. However, I do like the idea of taking the settings half way down so that I can simulate a higher dynamic range with less noise. Its the reason I asked my original question.
I am only talking about video because with photos, I do only shoot in RAW because I do like having the ability to "recover" an image if I screw up my settings a bit. But when it comes to video, I am not necessarily looking for tack sharp nor am I striving for the soft cinema look. I am currently just looking to see what everything will do. I just want the information to know what I have to do if I am ever called to do it.
Thanks for the information!
Well explained...why anyone would hit dislike on this is beyond me. Good job.
Brother for low light video which picture profile should I use???
Well, I have a Canon T6i and was wondering what can I do to take the best video shots out of it. Should I use an external downloaded flat profile or just customize the ones that came with the camera? Can you guys please help me out?
Tall Guy, I'd like you to run some test in Cinelike D using the Leeming Lut One, the last version is V502.
does this still apply to panasonic's HLG in 10-bit 4:2:2?
On my Lumix G7 I use the "Normal" picture profile and I have the contrast and sharpness brought all the way down. Much better for skin tones than the "Cinelike D" profile.
Same in my GH4
Yep, i do the same on the canon sx50 and also bring the saturation and colors down
Smart with my GH5
Wow, as i read the title i thought it would be some funny stuff about "you should absolutely blow out your Colors" or something, but this Video was really intersting and thoughtful! Thanks for that!
Glad you enjoyed it!
Very helpful thankyou very much!
Thanks, glad to help!
8 bit is a good point, but bit rate and the how the codec treats tonal range is a little important too. Compression affects banding faster than any else.
True, but generally a camera that records 10 or 12 bit color will have a more than adequate bitrate, even using just h.264. With higher end 8 bit cameras like an A7s or a GH4 your mileage will vary a lot compared to a lower end G7 or a6300.
Totally true. I have many people disliking the way VLog affects colors, Clog and Slog seem to work better. the 422 definitely helps, even in eight bit.
I think your overall message is valid and one that I agree with. However, there is a fundamental inaccuracy that you need to correct. In an 8 bit image, there are 256 potential values for all three color components: red, green, and blue. So there is a total of 256x256x256 potential values or 16.7 million possible colors.
You are correct! My bad. Thank you for pointing out that correction.
So basically with 8-bit you have 16 million colors and with 10-bit you have 1 billion. I have found banding is relatively rare in 8-bit images, since it is often masked by noise. (A little noise is your friend.) Problems come when you try to grade an 8-bit image. If you want to be able to work with footage in post, you need 10 bit. 8 bit is fine if you like what you are getting out of camera. But I agree: Log and 8-bit don't mix!
Actually it means, leave always at least 2bit headroom if you use a flat profile. If you want to deliver 8bit, using VLOG 10 bit is great. If you need to deliver 10bit (there are 10bit monitors and TVs out there), it might be tight with a GH5 and a flat profile. Thanks for this video!
That is true, you will want more headroom for delivering HDR content, but most people aren't delivering for HDR yet since the technology isn't very widely supported right now.
So what is your point? is this for the all in one video maker shooting on 8bit cameras?
Yep thats why i test all slow motion functions of my GH5 (in a video in this channel) and i saw weird colors transitions because of 8 bit and my grade...Thanks for the video you confirmed my thoughts ;)
Glad to help out.
I don't really have any issue with my camera C-log... or rather I don't really see any problems... I use a C100 mkii... I think I will stick with the Log profile just because I like to have a better flexibility with my image when I colour grade...
The C100 is a proper cinema camera, so I would say it does have the capability to properly record Log without any problems. The types of cameras I'm mainly referring to in this video are DSLR and Mirrorless cameras.
Now you're just being dishonest. The C100 MkII records at 8 bit, 35mbps 1080p. How can you argue that that is a stronger codec than for example a Panasonic G7 which records at 100Mbps 4k. Just because something is expensive and has cinema in it's name, doesn't mean that it can handle log profiles better. C100 has great low light, but when it comes to image quality it leaves much to be desired.
Great information right there.
If you want a video without banding although you make a color grading it´s recommendet to buy a camera with 10 - bit support.
Thanks for sharing, but I have to disagree. I film with a canon 6D and contrast presets take a lot of info away, so I use flat and it´s so much better, no matter if day or night. Im trying Technicolor cinestyle profile now, it´s harder to grade but I think it´s only matter of time to get it right..
I really appreciate this video Tall Guy, as it finally explains why it's bad to shoot flat video (something that pros do with super expensive cameras) with a prosumer dslr. A true eye opener.
Question though: If shooting flat is essentially bringing the values of the pixels more closely together, why did you recommend to do this in your G7 video by making the shadows lighter and the highlights darker? Wouldn't this be bringing the pixel color values more closely together?
Thanks.
Glad to help!
The way the "Highlight Shadow" curve control works is slightly different than a log picture profile. A log profile will change your absolute black and white points, while the curve only changes the contrast in the midtones. This does slightly reduce the amount of information in the midtones, but it preserves much more details in the shadows and highlights, so it's still a good idea.
Basically it's just a tradeoff.
Rex Loyer it's totally different that what pros do. pros shoot log or prores, our 8bit cameras can't output all the information theirs do. they don't shoot flat, the dynamic range is that wide that it is outputted as a flat image.
Dude thanks for that!!!! I experience serious problems with color corection and grading using this freaggin cinestyle.. what everyone is tellin me to use...
You need to use corrective lut for cinestyle.
I'm doing fine with Cinestyle
Hey @jas de guzman, Im still struggling here but I think maybe its a post production issue. What do you edit with? Feel free to DM me on IG at JTCFilms. Would love to chat more settings.
Where did you learn this??
Can someone help me ?
So can the canon 80d shoot falt or not
I dont know the bitrate cant find it
Thank you
80d can't shoot log. Check the properties of image for bit information.
I just thought I'd share my latest experience. I've got a gx85 with 14-140 pana lens and I just ruined my holiday videos with using a very flat profile. Somehow I ended up adjusting the shadows +5, highlights -5, I don't even remember when I did this. Even the contrast was turned down. I didn't see anything weird on the small camera screen but when I came home, watched the clips back on my pc I was shocked. The result is horrible. Huge amount of noise in dark parts even on skintones and highlights, and some of my videos don't even look 4K (they are though). Looks like if they were made by my 10 year old 1080i camera, or a cheaper phone. There's no way to improve the quality in post. The details aren't there and the noise is unbearable even in bright conditions. I made videos before with the same camera and lens but without touching anything in the menu, those ones look WAY better!
And I even underexposed my clips because I used a 100% zebra which showed clipping almost all the time. The ones which are overexposed look a lot sharper and less noisier.
I'm still not sure how to expose properly without loosing details in the highlights (sky with clouds) and not to underexpose. I think I could live with loosing details in the highlights but not with this kind of really ugly noise. (I forgot to mention my ISO was around 200 and 400 all the time.)
So from now on I will be more careful and I'll try to get the best looking results out of the camera and not to rely on post processing. I might lower the contrast and saturation in camera just a little bit because I like to way it looks. Sharpening and noise reduction will probably go all the way down, but I still need experimenting with these.
I shoot wedding everyweek with a A7S some times using slog2 i can say that all this guy is saying here IS TOTALLY TRUE all those dislikes are people puting in risk it's job.
Thanks mate for the video
Glad to help!
That Philmont shirt 🔥
noise ok is higher, but you can noise ninja and still have much less noise than from ML raw, and what about much higher Dynamic Range and most importantly color separation and color depth ?
I have a Panasonic g7. Your point seems to make sense but why does everyone recommend that you should shoot in a flat profile? Do you think there's any chance you could make a video shooting in a flat profile and a normal one and then colour grade both and see which is better? Thanks
Shooting in flat profiles gives you more dynamic range, which can be useful. Most people just don't consider the negatives. I might do a future video looking more in-depth at Log profiles.
pss try to color grading wihout the flat profile, its imposible, de image crash
I BELIEVE WHAT YOU SAID THESE SENSORS ON DSLR AREN'T REALLY BUILT FOR FLAT
It's also important to think about banding, having big gradients inyour video(aka a sky will. Make it much worse)
Nice explanation. Even someone like me who is completely new to videos unerstood.
And it is nice to hear the other side of the argument.
Clicky baity title though.
I definitely agree the title is a bit clickbaty, but unfortunately it's what you have to do on UA-cam sometimes. I'm glad you found the information in the video useful.
Yeah. Everyone does it. All cool. Content is more important. And you nailed it. :D
I'm having a problem refering LOG profiles as flat, as long as you properly expose a profile with a high dynamic range this won't be a problem at all. LOG is a really powerfull tool if you know what you´re doing. :)
Shooting flat picture profile is basically meant for professional videographers or should i say cinematographers and not for everyone who carries camera to shoot. I only shoot in flat color modes cos it gives me the room to play with my colors on post
Damn... where were you yesterday??? I just got into that EOS faking C-Log trap thinking I just stepped up my game. Well I've been wasting more than 6 hours trying to bring back those vivid colors. No more faking the pros. Anyway it's just a vlog. LOLL
lol
I do wedding videos and never ever shoot flat. I actually use vivid profile with contrast and sharpness at - 5 and saturation at 0
One thing people miss is that no matter what you shoot, whether 8 or 10 or 12bit, if you are just uploading to youtube, youtube compresses the video footage again.
That's true, which is why it's so important to consider your delivery destination when color grading a video.
This 8-bit analysis of flat picture profiles oversimplifies the encoding of image data in an H.264 file. While the individual luma and chroma components are stored at 8-bit (or perhaps 10-bit) precision, it is not simply a linear 8-bit RGB scale applied to the entire image. Instead, each frame is divided into 16-pixel square macroblock tiles and the average exposure level of each macroblock is recorded. This produces a logarithmic exposure scale that allows a much broader dynamic range to be encoded than a global 8-bit linear scale. Within each macroblock, 12-bit RAW RGB levels are converted into separate luma and chroma channels (YUV format) in the spatial frequency domain, and this data is compared to the average exposure level of the macroblock. The logarithmic difference is computed for each spatial frequency component and then truncated to a precision (quality level) determined by the bitrate allocated to that macroblock. It is the degree of macroblock component truncation (typically much less than 8-bit precision) that determines how much RGB color discrimination can be perceived in the decoded image.
When the H.264 file is loaded into a modern video editor, its image data is not converted into 8-bit RGB pixels. It is instead converted into an internal 32-bit floating point format, which allows lossless scaling and manipulation of image data. When you examine individual video frame pixels on a color monitor, the 8-bit RGB pixels you measure are not representative of the internal image data, they are merely truncated renditions scaled for display on an 8-bit monitor. When you save an edited video to a new H.264 file, image data is not encoded from the 8-bit pixels displayed on your monitor. It is instead rendered directly from the internal 32-bit floating point image data into 8-bit YUV format as described above. The 8-bit RGB pixels you see on-screen are never directly encoded into an H.264 file, that is merely a real-time rendering of the image data into a format suitable for visual display. The effective color resolution of the edited file is a combination of the video bitrates of both the original and edited H.264 encodings.
Log profiles are more intelligent at compressing dynamic range than just a linear compression (which is why they're called log profiles), which does help preserve detail in the midtones when using a properly designed profile, but that generally means that you have even less information captured in the shadows.
Once in an editing or color application you do have the full 32 bit float precision to work with, but that still doesn't give you access to information that wasn't captured in the original footage. The more drastic the color and contrast changes you make while editing, the more visible compression artifacts will be, especially while working with footage that isn't taking advantage of the full dynamic range of the original compressed file to begin with.
That's a related but separate issue. An ideal log profile maps each stop of exposure to a numerically equal slice of the linear 8 or 10-bit H.264 encoding range. A traditional Rec-709 profile rolls off the highlights and shadows, reserving more of the encoding range for midtones. In practice, log profiles also roll off the shadows but leave the highlights flat. But all this refers to how the image details look when decoded for display on a (Rec-709) monitor. Behind the scenes, the H.264 encoder converts the orginal RGB data into YUV spatial frequency components and records floating point delta values as I described above, rather than directly recording an 8 or 10-bit rendition of either log or Rec-709 profiles.
So broadly speaking, are you saying that you are getting more dynamic range at the expense of granularity when using a flat picture profile? So a flat picture profile will be disadvantageous in an evenly lit scene but can help in high contrast situations?
Essentially, yes, that's the TLDR. Unless you really need the additional dynamic range you're better off not compressing the image's dynamic range.
@@AlteraSound Thanks for the info, I appreciate it.
This was a really good nerdy explanation, and I agree and disagree. The reasons come down to how serious you are about filming if you really are, you should use flat because no matter what micro details this brings it helps you LEARN because of practice. and that is major KEY on becoming a good filmmaker. if you do this just for fun and hobby than it really doesn't matter. good quality video is very important in different ways but the story is still KING.
That totally depends on
1. How much you prioritise learning over creating the most professional result now (I watched this video to help me shoot clients + myself the best quality videos)
2. Sounds like you actually totally agree, you just like breaking codecs for fun to experience first hand how they break
Thanks for the video! Personally we record in slog3 8 bit for the last 2 years and achieve beautiful results with little or no issues in grading it. But that is also cause we shoot in 4K which holds up much better in post and acts more like 10 bit. Dave Dugdale has run some good video tests comparing 8 bit 4k against 10 bit and came to that conclusion. For us, we simply love grading footage to capture more dynamic range and options for LUTs etc. Having control over your image is paramount. I would never shoot a big job without it.
Awesome
I use the gh5... I am al Videographer and I think I have to learn some parts new :)
Useful
Is this inforrmation outdated? Canons entry level camera the M50 has a 14 bit DIGIC
That must be the name of the processor because I have the m50 and it does not record in any bit depth more than 8 bit. So idk where you got 14 bit from
The questions is... how to use histogram for DSLR?
The question is simple what you wanna do with it.. Just point and shoot? then leave it off. But if you planing something with colorgrading. You will need a flat profile. Every professional film cam uses flat profiles. Because of post production.
Then leave it to the professionals. 99% are "professionals" schooled on Facebook groups.
You have no clue what you are talking about. Professional films use RAW video which has no picture profile at all, and they grade from that. RAW has 12 or more bits of color latitude per pixel whereas your cheap junky Sony a5000 body shoots 8-bit compressed footage that is poorly suited to post-production grading BEFORE you totally screw up the contrast with your flat BS.
@@JodyBruchon how about Sony a7iii?
@@Emsyaz I don't use Sony cameras so I don't know everything about every model. If it can't explicitly output 10-bit files, don't shoot flat or log.
@@JodyBruchon how do I know whether or not a camera can explicitly output 10 bit file?
simply put. A flat profile should be used to bring out more details for flexibility in post Color grading.
IMO
I use flat when the shot has a lot of dynamic range or something’s Color is not very nice then I flatten it and regrade it.
If the shot is in a controlled studio then I would suggest shooting normally.
great explanation!
Thanks!
I’ve noticed that when grading log footage to proper blacks and whites, it offers no significant advantage in dynamic range compared to a standard in camera profile. Log is just a tool for specific situations.
I agree, it's a situational tool. Unfortunately too many people make more work for themselves by falling into the trap of believing you should always use flat or log profiles.
might wanna throw a little more scatter on that gradient there. bandy!
Or just dither it, which I've done in my more recent videos. Live and learn.
totally, i'd say never use flat PP in 8 bit color space!
What I do especially if I am not shooting with an ARRI and am going to be asked to make the same shot look bluish and dark and orange and dreamy bright so the director / art director can get a feel of what he wants to do with the shot then here is a useful idea. First you are most likely shooting 4:2:0 8 bit video with h264 encoding then use a Neutral profile or one that looks as close to what you are doing with the shot. Then tone balance (not white) to whatever you want to emphasize in the shot (like skin tone), the blue of a certain car, the variations of green in some forest wide shot. Then I would get the exposure where it would look like the mood - then I would shoot some stills and check out the shadows, then I would if they are clean go ahead shoot like that. Then you could just use it from camera or make very small adjustments in post - that's the way to work in 8 bit color. take care. Using a flat picture style will only makes things worse unless its at least 10 / 12 / 14 /or 16 bit color. Also if you flatten 8bit you have nothing to go from except flat footage and some stills - you will find banding issues in skys, you will have artifacts because the camera output is really not designed to be edited. Even flat footage will have less bits of color to work with - because the bits are all needed for a good image, if you use them to boost up your luma in the shadows (the idea behind flat) then all you are doing is pushing the colors that were in the middle up where they have less actual saturation like skin tones, then in post you bring them down but with out the subtlety you could have captured them in camera - so you add LUTs and false color to make that up. But if you compare very carefully shot 8bit in neutral settings to graded 8bit hands down most people will like the in-camera shot rather than the graded one. in blind tests of my own I found this to be true.
Since when it's about copy reality 1:1?
True! But color grading looks awful on non-flat shots. :/
I strongly disagree. I shoot using a flat profile on my Canon 550D (T2i) because the image simply looks better for the type of shots I mostly use (indoors, stage lighting). It makes my shots look more like they're from 5D.
He is not wrong. And you can use a picture profile that is not flat, AND it retains more dynamic range than the third party c-log, or even Technicolor Cinestyle. Look at Lightform C and Visioncolor CineTech. Both have equal DR, way more than Cinestyle, and they require little grading because they come properly right out of the camera. The least grading you do to 8 bit, the better it will be. If you can get the look you like right out of the camera, that's the best for 8 bit.
Eugenia Loli so let’s say I get the Canon M50, it would be best to shoot in the natural profile that comes with it, rather then in downloading the Cinestyle?
@@ManicMK More dynamic range with the Cinestyle. For me, that's the one reason to use flatter profiles.
Eugenia Loli thank you
I hate the number 256 because of video games.
Do you mean flat profile is log profile? you are right if ypo talk about log profile - there are issues with 8bit logs. But flat profile doesnt compress image.
aka- don't use log with 8-bit. Well explained -but that clickbait tile was a bit misleading.
Aye Scout brethren!
Indeed!
Preach it
Mr. Laa-Laa great name.
some people just don’t know how to color grade. They use luts but they not being honest lol
Just use common sense: generally it's preferred to shoot as flat AS POSSIBLE and as TGF mentions above in the video you will get color issues if you go flatter than your camera can actually process in a camera that doesn't shoot proper raw or log. At some point, if you're using a flat profile on a lower end camera you're not capturing AS MUCH information as you need for color, only for dynamic range. This is because unless you shoot true Raw or Log footage, properly white balancing your footage actually still matters. Otherwise you'll end up with pink and purple shadows and blue or green highlights... because again your flat profile isn't capturing the relevant color information you need. You should also look at polarizing filters if you need to get HDR but don't want to sacrifice color depth when shooting flat.
Nice tis