DSLR footage looks better on a better computer: Seriously.

[f64manray]

Very interesting, Kholi. After much research, I think I'm also going with CS5. What are your thoughts about editing the 5D/7D footage natively compared to transcoding. There's been some comments around the internet that transcoding to Pro Res results in a better image. I'm really hoping that editing natively will be just as good.


It seems like the only way to determine exactly what the DSLR footage truly looks like is to view it directly via HDMI cable to a large screen LCD. In any case, I've been testing the 5D and have always viewed the footage directly from the camera on a 52" Samsung LCD, and it has always looked impressive given that the lighting is there. I think the quality of lighting may also have something to do with the whole aliasing/moire thing as well. I have generally been testing in indirect natural light and haven't run into any problems anyway, but I'm not pushing it either. Soft light may provoke the sensor less than hard light for instance. This has been my experience and Laforet has also mentioned good lighting may play a role and Yataines seems to concur. You say the the DSLR looks better on a better computer. Could you expand on that? Has this reshaped your opinion on DSLR's?

 

[Kholi]

Kholi, you do this all on an 11" screen? i'm not even being sarcastic.

Haha. Definitely. It's a widescreen LCD and resolution's 1280x720 (actually I think it's 1366x768 max) so it's all good.

Also, again, there's a Display Port and HDMI out, so if I need another monitor it's all good.

 

[Kholi]

Very interesting, Kholi. After much research, I think I'm also going with CS5. What are your thoughts about editing the 5D/7D footage natively compared to transcoding. There's been some comments around the internet that transcoding to Pro Res results in a better image. I'm really hoping that editing natively will be just as good.

Someone got on my case a long time ago about saying Pro Res made footage look a bit LESS than the H.264. After seeing the 5DtoRGB plug-in, I feel validated enough to say that I wasn't seeing things.

CS5 allows you to edit DSLR footage (it even has PRESETS for DSLR Video, that's insanely awesome) right on the Timeline and rather efficiently, given that you can choose which resolution you edit at by simply right clicking on the video windows.

Being able to edit native H264/7D/5D footage just makes me want to shoot more, especially since you can also do it with RED footage (no converting RED to something less means SO many hours saved in getting to finally cutting.)

I recommend at least trying it if your workflow isn't based around FCP. For personal use, I can afford to get used to CS5, since it fits into my personal computing environment as of right now.

It seems like the only way to determine exactly what the DSLR footage truly looks like is to view it directly via HDMI cable to a large screen LCD. In any case, I've been testing the 5D and have always viewed the footage directly from the camera on a 52" Samsung LCD, and it has always looked impressive given that the lighting is there. I think the quality of lighting may also have something to do with the whole aliasing/moire thing as well. I have generally been testing in indirect natural light and haven't run into any problems anyway, but I'm not pushing it either. Soft light may provoke the sensor less than hard light for instance. This has been my experience and Laforet has also mentioned good lighting may play a role and Yataines seems to concur. You say the the DSLR looks better on a better computer. Could you expand on that? Has this reshaped your opinion on DSLR's?

That's how I first viewed DSLR footage, straight out of the camera via HDMI. For me, it was the only way to really tell what you were getting, even more so if you had a Production LCD to work with.

Certainly, the quality of light impacts the image. These are things that typically wouldn't matter much with film, but even with RED using an odd light source can be a problem. It's been a little easier with the MX upgrade, still you need to watch it.

It depends, I think, on the quality of the hard light. Shooting with Arri fixtures or Joker HMI's absolutely yields different results for me than using my Cool Lights HMI. However, it's what I can afford so I use it anyway.

For the record, I've always pretty much viewed footage as it was shot. It was a recent discovery for me that a lot of people may NOT have been viewing it at its best, though. I think I'm being misunderstood (with reason) about how I feel on the topic of DSLR video.

As a wanna-be DP, I'll always want the best I can get. There's nothing wrong with "good enough" not being good enough for me. However, I recognize--and don't think that I've ever contested--the benefit of having the 7D or 5D available to use when you can't grab better OR it fits the job.

Having learned the 7D and 5D inside out, and I say this plenty but not so much around the boards, I could and would confidently shoot a feature film with multiple cameras and not really worry about the shortcomings really impacting my production. That's because, of course, I know what to do with them, when to use them and when not to.

I still use them plenty and with great results. I just don't think they need to be compared to anything else.

 

[Rick Burnett]

First, you are absolutely right and a lot of people don't know that accelerated video cards are NOT all equal. I've noticed this before in the past. Especially in cases where the video card is being asked to scale images, those internal routines are NOT all equal. Heck, even the same video card and multiple displays you can see this sort of stuff. What I tell people if they cannot afford expensive gear or calibration is get some video YOU KNOW and play that one your system. Then you will see where the problems are and can *try* to fix them as best you can.

Pertaining to H.264 and ProRes, I hear a lot of people complain about Gamma issues, mostly with Final Cut Studio. Also, when you start pushing the H.264 around, also from what I have read, it doesn't behave as nicely. Not sure exactly what that means, but I've seen it written A LOT.

I use ProRes converting with mpegstreamclip because I put ALL my footage from whatever source to ProRes. It means I know what to expect from everything and that to me makes the whole process of editing, color correction, at least more known in what to expect.

 

[bwwd]

My whites are a bit more and faster blown out on my monitor than on my TV so im watching many things on my TV as this is where it should be displayed in the first place anyway.
Not all graphic cards are equal but its good that it looks ok at least on my TV,i even do color correction using TV because colours are more accurate.

 

[Kholi]

First, you are absolutely right and a lot of people don't know that accelerated video cards are NOT all equal. I've noticed this before in the past. Especially in cases where the video card is being asked to scale images, those internal routines are NOT all equal. Heck, even the same video card and multiple displays you can see this sort of stuff. What I tell people if they cannot afford expensive gear or calibration is get some video YOU KNOW and play that one your system. Then you will see where the problems are and can *try* to fix them as best you can.

Certainly true. I expect it to be more so lost knowledg on filmmakers without access to personal IT connections. Coming from the gaming world, it's always been imperative to have the latest drivers etc or your gaming would suffer. I'm silly for not thinkig of this before.

As far as correcting goes, any heavy grade is a cause for DPX sequences anyway. At least for me. Being able to cut the h264 and then export a picture lock in dpx is am easier workflow than converting then cutting.

At least for me. For tweaking an image, bumping contrast or gamma, saturatioN or white balance correction I think the h264 is plenty.

Pertaining to H.264 and ProRes, I hear a lot of people complain about Gamma issues, mostly with Final Cut Studio. Also, when you start pushing the H.264 around, also from what I have read, it doesn't behave as nicely. Not sure exactly what that means, but I've seen it written A LOT.

I use ProRes converting with mpegstreamclip because I put ALL my footage from whatever source to ProRes. It means I know what to expect from everything and that to me makes the whole process of editing, color correction, at least more known in what to expect.[/QUOTE]

 

[Ben_B]

Even if using an NLE that supports native editing you will always be better off converting to a codec like ProRes if you're going to be doing any serious color grading, etc...but really you're better off doing it no matter what because your computer will run faster, and you won't degrade footage as you edit...why edit in a long GOP codec? It causes lots of problems.

If you were really strapped for time/storage I suppose you could edit natively, media manage to ProRes, and then color grade...but again, not ideal.

Storage is cheap, and any time spent transcoding is worth every bit of piece of mind and improved speed while editing.

 

[Kholi]

Even if using an NLE that supports native editing you will always be better off converting to a codec like ProRes if you're going to be doing any serious color grading, etc...but really you're better off doing it no matter what because your computer will run faster, and you won't degrade footage as you edit...why edit in a long GOP codec? It causes lots of problems.

If you were really strapped for time/storage I suppose you could edit natively, media manage to ProRes, and then color grade...but again, not ideal.

Storage is cheap, and any time spent transcoding is worth every bit of piece of mind and improved speed while editing.

Computer will run faster? At this point, most new computer and GPU combos, at least at pinnacle build levels, can handle H.264 just fine. If there's no need to convert, why convert? That's an old way of thinking. Again, my M11X Plays back H.264 on the timeline without hitch, glitch or flaw @ Full Resolution, and the same piping it out of the HDMI port which needs even more processing power.

Converting is probably beneficial for lower end systems... sure.

None of this matters when you have a system that can handle whatever you throw at it.

It really has nothing to do with storage.

And why would you offline ProRes, anyway, if you're going to do any heavy grading? That's an awkward workflow... as ideally any serious grade would want ProRes 4:4:4:4 and not HQ, or better still DPX sequences.

So, you're going to do two conversions?

Again, this is old thinking for old computing hardware. There's nothing wrong with cutting the H264 right on a timeline if you have the ability to do so without hitch or glitch.

 

[Ryan Farnes]

Assuming my Macbook Pro were able to run the h.264 "without a hitch," isn't h.264 being used to bake any transitions, overlays or image/color corrections into the footage?

Yes, I use ProRes so my computer can run it smoother, but I also use it because if I don't, Final Cut is doing all that "baking" using a codec that loses detail every generation.

I was involved in a thread a few months ago that touched on this, but I never fully understood, is there a way to set ProRes as the "render codec" so that new data gets rendered into ProRes and upon final export, you're still dealing with pristine original h.264 footage and any new data/footage that was created exists as ProRes...thus resulting in no h.264 compression loss during rendering prior to final export?

Is that what "media manage" is?

 

[Kholi]

Assuming my Macbook Pro were able to run the h.264 "without a hitch," isn't h.264 being used to bake any transitions, overlays or image/color corrections into the footage?

Yes, I use ProRes so my computer can run it smoother, but I also use it because if I don't, Final Cut is doing all that "baking" using a codec that loses detail every generation.

I was involved in a thread a few months ago that touched on this, but I never fully understood, is there a way to set ProRes as the "render codec" so that new data gets rendered into ProRes and upon final export, you're still dealing with pristine original h.264 footage and any new data/footage that was created exists as ProRes...thus resulting in no h.264 compression loss during rendering prior to final export?

Is that what "media manage" is?

Technically, the process you're talking about is "offlining". Or, if someone wants to correct me?

But, you could output ProRes LT or even ProRes Proxy, keep the same file names as the original H264 then reconnect all of your media with the H264 files in a new sequence saved in a new location and you'll have a fullblown H264 online.

I used to do this with RED footage. Now, in Premiere, I just cut the R3D files/RAW red footage.

I'm not sure what would get "baked in" if you're just editing, though. You don't really do color correction and/or/then grading until you've locked your cut.

I suppose you could do a 1st light, depends on your process, though.

I think you could run H264 just fine at 1/2 res or 1/4 Res if you used Premiere. Again, it's well integrated for that NLE, which makes it a much more valid option than converting.

We had this same discussion for nearly a year when the HVX was released. Edius was one of the first NLEs able to cut raw MXF files, no conversion necessary. So, why convert if you have to?

You don't.

 

[Kholi]

Assuming my Macbook Pro were able to run the h.264 "without a hitch," isn't h.264 being used to bake any transitions, overlays or image/color corrections into the footage?

Yes, I use ProRes so my computer can run it smoother, but I also use it because if I don't, Final Cut is doing all that "baking" using a codec that loses detail every generation.

I was involved in a thread a few months ago that touched on this, but I never fully understood, is there a way to set ProRes as the "render codec" so that new data gets rendered into ProRes and upon final export, you're still dealing with pristine original h.264 footage and any new data/footage that was created exists as ProRes...thus resulting in no h.264 compression loss during rendering prior to final export?

Is that what "media manage" is?

Technically, the process you're talking about is "offlining". Or, if someone wants to correct me?

But, you could output ProRes LT or even ProRes Proxy, keep the same file names as the original H264 then reconnect all of your media with the H264 files in a new sequence saved in a new location and you'll have a fullblown H264 online.

I used to do this with RED footage. Now, in Premiere, I just cut the R3D files/RAW red footage.

I'm not sure what would get "baked in" if you're just editing, though. You don't really do color correction and/or/then grading until you've locked your cut.

I suppose you could do a 1st light, depends on your process, though.

I think you could run H264 just fine at 1/2 res or 1/4 Res if you used Premiere. Again, it's well integrated for that NLE, which makes it a much more valid option than converting.

We had this same discussion for nearly a year when the HVX was released. Edius was one of the first NLEs able to cut raw MXF files, no conversion necessary. So, why convert if you have to?

You don't. Not until you're ready to go to the next stage.

 

[Rick Burnett]

I hardly think comparing R3D to H264 makes sense. You are talking about a storage formate that stores MORE data compared to one that stores LESS.

I have NEVER heard the argument that working in H264 was about processing power. My machines have been able to handle H264 for quite awhile. What I heard, if you are doing hand-offs between different applications, using H264 as your format means you could be throwing away useful data that might get brought up in other processes. Storing in a higher data depth where you go back to file between possible operations means more retained information.

Now, if you work with the H.264 all in one program so that internally it isn't handled at the lower rates, I could understand that.

It's the same thing with audio. If you are using multiple audio programs, storing as a 24-bit file truncates extra detail that you might have had inside at 32-bit float. This is why I always try to do all my audio editing inside one app so things like low pass filters don't add in a ton of quantization error. (I come from an audio background).