EVA1: Music video shot in a couple of hours, with zero budget, and only a vague plan
- Thread starter Seanik
- Start date
Bassman2003
Veteran
Nice images. Liked it. Thanks for sharing. How low is zero budget when an $8,000+ camera is involved along with talented shooter? 
Unless it was a favor.
Unless it was a favor.Thank you!
Zero budget - yeah, it was a favor/gift (my friend the singer had just earned her PhD)
- used only available light (ok, I used a little bounce on the CU shot with the flowers... and should have used the 3x4 silk/reflector, but I couldn't monitor for shadows and movement caused by the breeze)
- b-roll shot at 120 fps, and singing sequences shot at 42 fps (if I recall correctly). Interesting live/slo mo effect (when it's synched to the audio correctly, that is)
- Vlog -> ACES -> Rec709 for grading in Resolve
Boy, did I learn how much Vimeo and YouTube conspire to undermine a lot of grading work, lol! (e.g. I used Resolve's "Color Compressor" effect to conform the main hues toward a specific color palette... looks great when exporting... even the h.264 looked good... but horrible when uploaded and recompressed!). I also learned that just because the camera doesn't produce banding on gradients such as the sky, it doesn't necessarily mean that Vimeo or YT will let you get away with it. ¯\_(ツ)_/¯
Fun to work on, even if it was all semi run-n-gun. The EVA1 really shined here... the DR alone is what made this project even remotely possible. This was also the first time I shot with it where there was an opportunity to capture those kinds of colors. Just, wow. It's amazing how brilliant and rich it looks without it being oversaturated and plastic.
I'm not sure I could have charged anyone for this and felt good about it, though Thanks for having a look!
Zero budget - yeah, it was a favor/gift (my friend the singer had just earned her PhD)
- used only available light (ok, I used a little bounce on the CU shot with the flowers... and should have used the 3x4 silk/reflector, but I couldn't monitor for shadows and movement caused by the breeze)
- b-roll shot at 120 fps, and singing sequences shot at 42 fps (if I recall correctly). Interesting live/slo mo effect (when it's synched to the audio correctly, that is)
- Vlog -> ACES -> Rec709 for grading in Resolve
Boy, did I learn how much Vimeo and YouTube conspire to undermine a lot of grading work, lol! (e.g. I used Resolve's "Color Compressor" effect to conform the main hues toward a specific color palette... looks great when exporting... even the h.264 looked good... but horrible when uploaded and recompressed!). I also learned that just because the camera doesn't produce banding on gradients such as the sky, it doesn't necessarily mean that Vimeo or YT will let you get away with it. ¯\_(ツ)_/¯
Fun to work on, even if it was all semi run-n-gun. The EVA1 really shined here... the DR alone is what made this project even remotely possible. This was also the first time I shot with it where there was an opportunity to capture those kinds of colors. Just, wow. It's amazing how brilliant and rich it looks without it being oversaturated and plastic.
I'm not sure I could have charged anyone for this and felt good about it, though
Thanks for having a look!
Last edited:
Bassman2003
Veteran
Heck ya you could have charged for this. It all has some level of value. It is a shame that YouTube has to equalize everything in a time where cameras are making exceptional images.
Clermond
Veteran
Nice images but for my taste you could have kept the singing part in color and the „acting“ part in black and white or vice versa. There is also no need for a strict ACES REC709 colors imho, maybe kind of subtle warm look would look nice in contrast to the B&W scenes. No offense, just my personal taste.I watched it till the end which is definitely a compliment .
Last edited:
Alex Derrick
Active member
Have you tried exporting from resolve in Quicktime DNxHD 444 and then upload that huge file into youtube. In my experience this is how to get the best results from YouTube compression by feeding it a file with lossless compression. In some cases even the varicam fine sensor texture can make it thru the compression algorithm when viewing in 4k or 2k.
Just a thought.
Just a thought.
Alex - Thanks for the tip. When I first started uploading and noticing these issues, I tried something similar by exporting a ProRes 4444 version, but it yielded the same results. I just tried both DNxHD RGB 444 10-b 1080p and DNxHR RGB 444 10-b on Vimeo, per your suggestion. (Because the source is 2048x1080, and the DNxHD codec conforms it to 1080x1920, I thought I would also try a DNxHR version that didn't mess with the frame size). Anyway, no luck... same artifacts, banding, color, etc.
For this final version linked and posted above, I used a few of the usual tricks (adding film grain, etc.), but ultimately I just had to back off the vignetting and lighten up the sky so the gradients weren't so aggressive. It is what it is, I guess. The lesson here is to minimize on shooting the sky (or maybe flatten it out a bit when grading), and don't expect to be able to use heavy vignettes (which I try to avoid anyway... this project was just a special case).
Rob - I used two lenses on this: Sigma Art 50mm f/1.4 and Tokina 11-16mm DX f/2.8. The Sigma, I thought, really looked great and is such a pleasure to use, always. The Tokina, for all it's flaws, is just fun to go hand-held and run about at f/5.6 or more (especially with the EVA1 internal stabilizer, it really "absorbs" a lot of extra shake). Bonus: when you grade to greyscale, you don't have to worry about CA's with the wide lens, lol. I don't think I opened up any more than f/3-4ish on either lens, btw (uber shallow DOF is not really my style... I like to see my subjects in focus once in awhile
).
And to everyone, big thanks for the responses. I really appreciate the feedback. Hat tip to Clermond for sticking it out to the end, lol.
For this final version linked and posted above, I used a few of the usual tricks (adding film grain, etc.), but ultimately I just had to back off the vignetting and lighten up the sky so the gradients weren't so aggressive. It is what it is, I guess. The lesson here is to minimize on shooting the sky (or maybe flatten it out a bit when grading), and don't expect to be able to use heavy vignettes (which I try to avoid anyway... this project was just a special case).
Rob - I used two lenses on this: Sigma Art 50mm f/1.4 and Tokina 11-16mm DX f/2.8. The Sigma, I thought, really looked great and is such a pleasure to use, always. The Tokina, for all it's flaws, is just fun to go hand-held and run about at f/5.6 or more (especially with the EVA1 internal stabilizer, it really "absorbs" a lot of extra shake). Bonus: when you grade to greyscale, you don't have to worry about CA's with the wide lens, lol. I don't think I opened up any more than f/3-4ish on either lens, btw (uber shallow DOF is not really my style... I like to see my subjects in focus once in awhile
).And to everyone, big thanks for the responses. I really appreciate the feedback. Hat tip to Clermond for sticking it out to the end, lol.
Clermond - I know it's not necessary, but I like the ACES transform from VLog much better than the V709 LUT. Maybe it's purely psychological, but operating in that color space seems much less rigid. If it was a quick turn, for sure I would just use Lumetri and the LUT (and maybe a "look" LUT).
In Resolve, it's just as easy, if not easier, to apply. Also nice to have an ACES master for later.
In Resolve, it's just as easy, if not easier, to apply. Also nice to have an ACES master for later.
I see ACES as a means, not (necessarily) an end, but ok. The only prohibition it presents to me is using someone else's out-of-the-box "look" LUTs (which is fine, if that's what turns you on. There's not much distro of interesting LMTs, to be sure, so finding your own way is up to you). My feeling is that, among a slew of other things, this was not the challenge here.
Erik Wittbusch
Well-known member
FYI: I never use LUTs or ACES for my EVA1 projects.
Try the Color Space Transform ResolveFX - it's the most flexible, node-based workflow.
Try the Color Space Transform ResolveFX - it's the most flexible, node-based workflow.
Maybe you guys realize this (maybe not), but let me just throw this out there, nonetheless:
In Resolve, I usually just set up the entire project in ACES:
Project Settings/Color Management
/Color Science = ACES [I just use ACEScc]
/ACES Version = [I just use the latest, v1.1]
/ACES Input Device Transform = Panasonic V35
/ACES Output Device Transform = Rec709 [or whatever you want to monitor and output to]
/Process Node LUTs in = ACEScc AP1 Timeline Space [or you can use AP0 Linear if you need it]
That's it. Done.
All of the clips on the timeline are now transformed to ACES (the colorspace that I'm working in) and output to Rec709 (the space that I'm setting levels for). There is no need to apply any transforms/LUTs in the node tree for each clip. It's a wholesale transform in one single step.
When working on a project with a mix of clips from different sources/colorspaces/gammas (i.e. VLog, SLog3, rec709, etc.), I do the same as above except for "ACES Input Device Transform" I select "No Input Transform." This simply changes the timeline to ACES and output to Rec709 without transforming any clips.
Next, I just go to the Media Pool, create new bins for each different source (e.g. VLog, SLog3, etc.), and add the clips to their respective bins (doing this does not effect their locations on your HD... it's strictly within Resolve).
Now, open one of the bins, select all of the clips, right click to get the menu, select "ACES Input Transform", and apply the transform of your choice (e.g. "Panasonic V35" , "Sony SLog3 SGamut3Cine", etc.). Do this for each bin of different sources. *For clips that are already Rec709, just select "Rec709" even if you are monitoring/outputing to Rec709... it functions sort of like a bypass, but really the transform goes like Rec709 [source] -> ACES [colorspace] -> Rec709 [monitor/output]... you're still really working in ACES even though it looks unchanged.
That's it. Done. The working space for your project is ACES and all of your transforms are already applied. No need to do the transforms in the node tree.
To take it even further, you can even create LUTs to match up sources (e.g. white balance, color), and use this same method to apply them to all the specific clips that you need to affect in one step. If you shoot a few frames of a chip chart with each different camera in each different scene, this is amazingly easy. One less operation that you have to repeat over and over again when grading.
This is my SOP when working with EVA1 on multicam shoots/projects (because chances are, I'm the only one with a Panasonic!).
In Resolve, I usually just set up the entire project in ACES:
Project Settings/Color Management
/Color Science = ACES [I just use ACEScc]
/ACES Version = [I just use the latest, v1.1]
/ACES Input Device Transform = Panasonic V35
/ACES Output Device Transform = Rec709 [or whatever you want to monitor and output to]
/Process Node LUTs in = ACEScc AP1 Timeline Space [or you can use AP0 Linear if you need it]
That's it. Done.
All of the clips on the timeline are now transformed to ACES (the colorspace that I'm working in) and output to Rec709 (the space that I'm setting levels for). There is no need to apply any transforms/LUTs in the node tree for each clip. It's a wholesale transform in one single step.
When working on a project with a mix of clips from different sources/colorspaces/gammas (i.e. VLog, SLog3, rec709, etc.), I do the same as above except for "ACES Input Device Transform" I select "No Input Transform." This simply changes the timeline to ACES and output to Rec709 without transforming any clips.
Next, I just go to the Media Pool, create new bins for each different source (e.g. VLog, SLog3, etc.), and add the clips to their respective bins (doing this does not effect their locations on your HD... it's strictly within Resolve).
Now, open one of the bins, select all of the clips, right click to get the menu, select "ACES Input Transform", and apply the transform of your choice (e.g. "Panasonic V35" , "Sony SLog3 SGamut3Cine", etc.). Do this for each bin of different sources. *For clips that are already Rec709, just select "Rec709" even if you are monitoring/outputing to Rec709... it functions sort of like a bypass, but really the transform goes like Rec709 [source] -> ACES [colorspace] -> Rec709 [monitor/output]... you're still really working in ACES even though it looks unchanged.
That's it. Done. The working space for your project is ACES and all of your transforms are already applied. No need to do the transforms in the node tree.
To take it even further, you can even create LUTs to match up sources (e.g. white balance, color), and use this same method to apply them to all the specific clips that you need to affect in one step. If you shoot a few frames of a chip chart with each different camera in each different scene, this is amazingly easy. One less operation that you have to repeat over and over again when grading.
This is my SOP when working with EVA1 on multicam shoots/projects (because chances are, I'm the only one with a Panasonic!).
Clermond
Veteran
Seanik, I was not talking about technical things. Is the ACES workflow correct? Yes. Does the image look correct? Yes. Does the correct image look appealing in this case? No offense, but not for my eyes. It's neutral, it's too clean. It's a music video not an interview. Just my humble opinion.
Also, there probably isn't really that much of a difference (for a strictly or majority VLog sourced project) to instead use "DaVinci YRGB Color Managed" for the Color Science, work instead in "Panasonic V-Gamut/VLog" for Timeline Color Space, and output to Rec709... V-Gamut is nearly as wide as ACES, after all. (I just have these ACES settings saved as presets, so I just go with what I know works well). I'm sure the two color sciences have their own pros/cons respectively, but I haven't really dug into that.
I do find, however, that there is a difference in working (grading) on clips that are in a wide gamut color space before the transform to Rec709, as opposed to grading after the transform step.
Anyway, hopefully someone may find some of this stuff useful. Thanks!
I do find, however, that there is a difference in working (grading) on clips that are in a wide gamut color space before the transform to Rec709, as opposed to grading after the transform step.
Anyway, hopefully someone may find some of this stuff useful. Thanks!
Clermond - I hear you my friend. I'm not offended at all (actually, I'm very grateful for your input). I also believe that I understand what you're saying: that the "look" is too neutral for a music video. What I am saying is that is not so much a function of using ACES, but rather my own (so-called) creative decisions and to some extent the re-compression by Vimeo and YouTube.
BTW, here's a link to an h.264 version of it before it hit the Vimeo/YT machine (it's quite a bit different). Better to download it and view it in QT or VLC.
https://drive.google.com/file/d/126sx76DiOxIS61UdHR-GQq0pFW_Pc-Lg/view?usp=sharing
I take your point about the neutral look. I actually experimented with a "technicolor 3 strip" technique, but found it too cartoonish and some of the colors were way off (including the skin tones). Instead, I opted to develop a color pallet for each scene and used the Color Compressor effect to "bend" the hues toward respective target hues that I sourced from some color theory charts. Interesting effect in that it actually limits the amount of colors but unifies them and sets them against their natural compliments. Alas, Vimeo/YT killed that, lol! But yes, you are correct that, on balance, I took a moderate to conservative approach on the color scenes. My goal was to produce bright and vibrant color (to contrast with the monochrome shots) without being too gaudy. The end result was that I was never in danger of that happening, lol! Again, thanks for the advice... on something like this, I could stand to push it a bit more. Honestly, I think I just fell for the colors right out of the camera, and didn't want to give them up!
Cheers!
BTW, here's a link to an h.264 version of it before it hit the Vimeo/YT machine (it's quite a bit different). Better to download it and view it in QT or VLC.
https://drive.google.com/file/d/126sx76DiOxIS61UdHR-GQq0pFW_Pc-Lg/view?usp=sharing
I take your point about the neutral look. I actually experimented with a "technicolor 3 strip" technique, but found it too cartoonish and some of the colors were way off (including the skin tones). Instead, I opted to develop a color pallet for each scene and used the Color Compressor effect to "bend" the hues toward respective target hues that I sourced from some color theory charts. Interesting effect in that it actually limits the amount of colors but unifies them and sets them against their natural compliments. Alas, Vimeo/YT killed that, lol! But yes, you are correct that, on balance, I took a moderate to conservative approach on the color scenes. My goal was to produce bright and vibrant color (to contrast with the monochrome shots) without being too gaudy. The end result was that I was never in danger of that happening, lol! Again, thanks for the advice... on something like this, I could stand to push it a bit more. Honestly, I think I just fell for the colors right out of the camera, and didn't want to give them up!
Cheers!
Last edited:
Great job on the music video! The creative of B&W with color technique... makes the colors pop even more. Nice job. My personal mantra of "taking something that's real, make it look not real, then make it real again" is what I think Clermond was referring to in regards to a more stylistic approach in the music video genre. Ironically, the fact that it is "clean" makes it stand out. Go figure! Anyways, that EVA1 (with your grading), is look'n mighty nice.