Question on how to approach this situation properly

I just set up a quick example in the timeline (screenshot) to illustrate my question.

What is the best way to approach this situation?

The cut clips on the timeline are from not very good quality 1080p longer videos (of 20-30 mins, or more), on a 4k timeline. Some with transitions, others with maybe with color corrections, etc.

(TL;DR, it's like using proxies the other way around I guess... without coming back to the original low res videos afterward)

What is the best way for replacing them with a better version of themselves?

I use another piece of software to clean up bad quality videos, but it's very, very slow, and I can't use it on the original non-cut video files, then use them do my cuts and edits... because it would take weeks to render them all cleaned up, so I would like to be able to cut the "original" low res ones (like in the example), and replace just the cut parts by their high res cleaned up version.

I was thinking about using the "render in place" option, clean up the exported files, then replace the ones on the disk with the new ones (with the same names).

Is that a good idea? Is there another way to do this properly and/or more easily?

I could cut all the part needed, export them as "individual clips" from the deliver page (but the naming scheme and with so many small files, it would take too much time to replace them one by one on the timeline, and find their right place)

Thanks

Resolve has several different methods for relinking multiple clips to new versions, so your worry about exporting individual clips is not a serious problem, I think.

Version 17 (beta) explicitly lets you create, export, and relink Proxy media. (This is explicit media that is meant to share with others, not the Version 16 timeline proxy media, which is not meant to ever explicitly be used or shared.) If you're on Version 17 already. Remember that Version 17 will want to update your database and once it does, you will no longer be able to open it in Version 16, so backup your DB and take precautions if the new Proxy media seems like a solution to your problem.

But even in 16 I believe there are options to Relink Clips for Selected Bins in the Media page, so if your timeline uses only clips from a particular bin, it's easy to relink them all. I believe they need to have the same names as the original files and be the same length, and if so, Resolve uses them and they magically update in the timeline as appropriate.

You should not have to use any method where you have to individually handle clips in the Media page or in a timeline.

I'd save yourself the hassle and try the tools in Resolve to improve the footage. What are you using externally?

I'd save yourself the hassle and try the tools in Resolve to improve the footage. What are you using externally?

It's not an option in my case because it doesn't do what Video Enhance AI does. The Topaz Lab software takes 0.35 up to 0.90 secs for a frame to render (depending on the hardware). It's super intensive. But the results are way better than what SuperScale can do (and I understand why they don't do the same thing anyway).

Resolve has several different methods for relinking multiple clips to new versions, so your worry about exporting individual clips is not a serious problem, I think.

I'm not worried, I'm asking because I never had to create proxies. I'm mostly looking for people that encountered this particular case, who are very familiar with Resolve, and want to share their workflow, what they are doing to accomplish what I'm trying to do.

I think my situation was misunderstood (maybe).

In this particular case, it's different (I think). I'm not trying to use proxies, then having Resolve uses the "full length" original files for the renders, but the other way around. I would like to export all the little cuts (dozens and dozens of them, from long video files) of the timeline, work on them externally, and replace them. Just the cuts.

That's why I was asking if the "render in place" wouldn't be the better option. I could export all the cuts in a specific folder, work on them with my external tools, and when Resolve is launched again, it would just see the same name of these "Render in place" files in the same folder, but the modified version. And I wouldn't have to relink anything (I guess?)

.....OR if there is a better way to go about this, a better workflow, or ideas from people who have WAY more experience with Resolve than me.

Ah, I'd understood you, but then forgot a step. When Resolve relinks it's looking for the entire original clip, which it can then conform in the timeline, and that's not what you have.

I'm not sure that "Render in Place" is meant to have its results accessible to users. Similar to the v16 Timeline Proxy files, which were never meant to be user-accessible. But I could be wrong.

That's why v17's Proxy mechanism is different: you are free to access the results.

I do think that File -> Media Management... -> Timelines -> Copy Used Media should do the trick. That should create a copy of your timeline and your media in the folder of your choosing. Then you can poke into the media folder with Topaz and process everything -- the "base" clips that you originally imported into your Media Pool are trimmed versions that only cover what's used in your timeline.

I just tried this approach on a project I have. I copied a timeline to a folder I made for that purpose and in that folder appears the timeline (.drt file) and the clips used in the timeline. If you import the .drt file, it's linked to the files in that folder.

I'm thinking if you have Topaz process the movies in that folder and output to a subfolder by the same file names, you could then do the Relink to those. The copy has made the Media Pool files and the clips on the timeline match in offset and duration.

Sorry I haven't done it and can't say for sure.

Trim down the clips to the parts you want to use. (Be sure to include some handles for transitions.) Don't do any other work yet. No color, no effects, no audio.

Export out those clip segments using the Individual clips option on the Deliver page to a decent codec like Cineform or DNx.

Process those new clips in Video Enhance, bring the results into Resolve for the real editing.

Trim down the clips to the parts you want to use. (Be sure to include some handles for transitions.) Don't do any other work yet. No color, no effects, no audio.

Export out those clip segments using the Individual clips option on the Deliver page to a decent codec like Cineform or DNx.

Process those new clips in Video Enhance, bring the results into Resolve for the real editing.

Thanks

I guess it's the only way to do it. The thing is, my example has just 7 of them and was just a way to show the context of my situation, but I often have 50 to 100 on average.

I'll try it on a small project (and try the other method I was thinking about, the "render in place" and physically replace the parts by the new ones).

We'll see.

Thanks for all the responses I appreciate.

I would still say the Media Management approach could work too, but I've experienced what I think is a bug in v17b6 when trying to do what I recommended. So for now I wouldn't do it that way.

I just set up a quick example in the timeline (screenshot) to illustrate my question. What is the best way to approach this situation? The cut clips on the timeline are from not very good quality 1080p longer videos (of 20-30 mins, or more), on a 4k timeline. Some with transitions, others with maybe with color corrections, etc.

What specific kinds of source material are you dealing with? What do you ultimately need to deliver?

One tough issue with workflow is that bad decisions tend to roll down hill, so making a bad decision early on to use low-res files will affect everybody from that point on (color, VFX, editing). It's critical to make the right decision, and have the ability to swap out the original files whenever necessary.

I once ran into a music video where they handed me an XML and the original camera files, and I realized to my horror that they had cut the whole thing in 29.97, but shot in 23.98. It was a huge load of grief, but I manually rolled up my sleeves and just dropped in the original 23.98 camera files in a new 23.98 session, because it looked that much better. It took some time, it was a lotta work, but ultimately the project benefitted from "doing it the right way." Could I have just lived with it and worked with those files? I could have, but that's not who I am.

I just set up a quick example in the timeline (screenshot) to illustrate my question. What is the best way to approach this situation? The cut clips on the timeline are from not very good quality 1080p longer videos (of 20-30 mins, or more), on a 4k timeline. Some with transitions, others with maybe with color corrections, etc.

What specific kinds of source material are you dealing with? What do you ultimately need to deliver?

Not a TL;DR.

If often have to deal with already rendered and highly compressed video files, mixed with "high quality" 4k files straight out of cameras/phones/whatever source. I always ask beforehand if they have good quality files, or if they can film something else. It can be for a "serious" project, or not. Most of the time there is no other choice than just work with what they (or I) have on hand.

...on 4k timelines. It's all about to "how to replace tons of little -cut parts- on a timeline without having to deal with the full-lenght original video files"

Until recently, I just told people that I cannot do miracles with what they are giving me, so it's not really a problem to have suddenly a pixelated mess (relatively speaking) in the middle of a clean 4k video. Even if I try to minimize it the best as I can, I can't invent details that are not there. That's where the proper upscale/de-blocking/cleaning up comes in.

The problem arose when I took more (personal) time on a project, cleaned up all the 720p/1080p clips. Upscaled to 4k, de-blocked, etc. Just to see what's possible to achieve. It took sooooo much time, I didn't have any proper workflow. Big mistake on my part ahah. I was asked why the other projects didn't have this level of "quality". I shoot myself in the foot, but fortunately for me, most of the people understand when you explain the situation.

Amateur or pro, personnal projects or multi-millions dollars feature films, I'm sure we all prefer to work with the best sources file possible.

...and sometimes the main video is a 1080p/720p very compressed file. When it's the case, and if the video is less than 20 mins, I can schedule the hours needed to render a cleaned-up upscaled version of it, and do the same manually for the 5-10 secs clips I will insert in the edit (if needed).

BUT, very often, the files I have to deal with are 30-60 mins long, I have 20, 30, or more cuts from 5-10 of 20-30 mins video files inserted on the timeline. It would take days 24/7 for me to render every single original files before I could start any edit.

I'm an amateur in this sea of pros, so I try to get the most tips and tricks possible

I will certainly update this thread if I find the best workflow for this particular situation.

I'd suggest to test this workflow:
1. Do the edit with the quality as it is, maybe after making sure with short tests if the material can be brought to point where it's not a reject. Trim the clips to be treated with the help of Media Management. It works well for me in 16, maybe there are still bugs in 17. Make sure to define subclips if you use more than one piece of a long source.
2. Treat the trimmed down material with VEAI, exporting high-quality clips.
3. Relink those into your edit.

I'd suggest to test this workflow:
1. Do the edit with the quality as it is, maybe after making sure with short tests if the material can be brought to point where it's not a reject. Trim the clips to be treated with the help of Media Management. It works well for me in 16, maybe there are still bugs in 17. Make sure to define subclips if you use more than one piece of a long source.
2. Treat the trimmed down material with VEAI, exporting high-quality clips.
3. Relink those into your edit.

Thank you, I'll try it the next time I have to do this

I'm not sure if clips out of VEAI preserve TC.
The software is really interesting for upscaling, but the last time I looked these guys were still learning about proper video.
So, better test it.

...on 4k timelines. It's all about to "how to replace tons of little -cut parts- on a timeline without having to deal with the full-length original video files"

I think the answer is a series of steps:

1) make sure you have enough storage before starting a project

2) convert what you need to convert to low-res proxies (which can be done easily in Resolve 17), making sure the filenames are identical.

3) edit with your proxies and lock the edit.

4) relink to your original high-res files.

5) use Media Management to create smaller copies of files (with handles) to a new drive.

All of this is covered in the manual. Now, you can deliver a high quality master without any compromises. You may run into some challenges if you have speed changes, complex effects, or non-timecoded material, but this method can work to a point. Worst-case, you could use Render in Place in Resolve 17 and see if that helps.

(BTW, you never did answer my question as to what the original source files are, and that has a huge bearing on workflow as well.)

...on 4k timelines. It's all about to "how to replace tons of little -cut parts- on a timeline without having to deal with the full-length original video files"

I think the answer is a series of steps:

1) make sure you have enough storage before starting a project

2) convert what you need to convert to low-res proxies (which can be done easily in Resolve 17), making sure the filenames are identical.

3) edit with your proxies and lock the edit.

4) relink to your original high-res files.

5) use Media Management to create smaller copies of files (with handles) to a new drive.

All of this is covered in the manual. Now, you can deliver a high quality master without any compromises. You may run into some challenges if you have speed changes, complex effects, or non-timecoded material, but this method can work to a point. Worst-case, you could use Render in Place in Resolve 17 and see if that helps.

(BTW, you never did answer my question as to what the original source files are, and that has a huge bearing on workflow as well.)

Thanks for the tips.

Mm, I think I did? -- "If often have to deal with already rendered and highly compressed video files".
It means 720p videos that have no other (and better) copies, 1080p videos that have made a couple of trips in the wrong hands, compressed again, etc. The last time I had to include parts an old video they posted on youtube, some 720p video uploaded 5 years ago. You can imagine how that looks. That is the extreme examples btw. The type of file and/or content doesn't really matter. But I care about the visual quality.

By default, I just take what people give me to work with, and I tell them how good or bad it can look like. The whole scenario of my post is because I want to try to do it more often, and more quickly.

I'd suggest to test this workflow:
1. Do the edit with the quality as it is, maybe after making sure with short tests if the material can be brought to point where it's not a reject. Trim the clips to be treated with the help of Media Management. It works well for me in 16, maybe there are still bugs in 17. Make sure to define subclips if you use more than one piece of a long source.
2. Treat the trimmed down material with VEAI, exporting high-quality clips.
3. Relink those into your edit.

So, I finally had the time to experiment a bit with that (on a very small "test" project). I didn't know that it could just trim the clips and save them as new video files without re-encoding anything (I used FFMPEG before that to cut parts of long videos without re-encoding), it's GREAT! It will save me so much time and hassle!

I had to learn a bit more about what's "conforming" is, because I'm not completely familiar with all to vocabulary around video editing. I like discovering new aspects of Resolve every day. And it helps to know what terms to search in the 3560 pages manual I tried many times to find responses in it, but not knowing how some things are called is very frustrating.

Anyway, for this little test, it worked well! I put all the cut clips in a new folder, did the VEAI pass, reimported in a new bin, then right-clicked on my timeline, conform from bins, selected the bin with the new files (I kept the same names), and it worked!

I'll test that too when I can, but if you already have the answer don't hesitate : If I make a Fusion comp with a clip on the timeline, does it get replaced with the conform method? - EDIT : OMG it does! So if I fix the output of the fusion comp to 4k, I can work on 1080p clips with fusion, and just swap them afterward.

Another question, I saw in the Conform Options : File Name > User Defined.
Is it possible to tell Resolve to look for the original filename + the extra added by VEAI? Like : original-filename.mp4 > original-filename-4k-upscale.mp4?

Did you check "Using Metadata to Define Clip Names" on page 333 in the March 2020 manual?
And "Available Variables in DaVinci Resolve" on page 3278.

Did you check "Using Metadata to Define Clip Names" on page 333 in the March 2020 manual?
And "Available Variables in DaVinci Resolve" on page 3278.

It doesn't give any option when writing "%" in the Conform Options > Pattern, or "User Defined"
I found something (I'm not really into "regex") - Chapter 47 - Conforming and Relinking Clips - p. 906.

In my case with the example that i gave would be : %D-4k-upscale.mp4 (or %_R, I'll test both later anyway)

So for those who are finding this page from google (I guess I'm in the "exotic" workflow ahah) :

-------------
Using the Pattern Field

If you’re using the Pattern option to extract the reel name from a clip’s source file pathname, you
have the option to create your own search pattern, enabling you to have DaVinci Resolve
extract the reel name in highly specific ways to accommodate more exotic workflows.

Extraction patterns are interpreted from right to left, deciphering each clip’s file path element by
element starting with the file name, and then considering each enclosing directory’s name to
the left.

Each extraction pattern consists of a series of text characters and “wild card” operators
in unique combinations corresponding to the length and names used in the file path.
Here are a series of search characters that may be used.
Extraction Pattern Operators

1. ? Looks for matches of any single character. Add as many question marks as there are
   characters you want to match. ?? matches two characters such as 02; ???? matches
   four characters such as 0002.
   
2. * A wildcard that creates matches for any sequence of zero or more characters.
   %R Specifies the reel name’s actual location. Reel names may contain any character, but
   should not contain a directory separator (forward slash).
   
3. %_R Extracts the reel name and strips out the R3D file name underscores found in EDLs
   from Final Cut Pro 7 or earlier.
   %D Matches any directory name or file name. When this is the last operator in a pattern,
   do not add a forward slash.
4. / Used to separate any two operators

Mm, I think I did? -- "If often have to deal with already rendered and highly compressed video files". It means 720p videos that have no other (and better) copies, 1080p videos that have made a couple of trips in the wrong hands, compressed again, etc.

I was looking for a codec type, which is a key factor in workflow. What you describe is a hopeless situation. If you have 72P H.264 material that's been compressed twice, there's nothing you can do. I would give up and live with it, or I'd go to the ends of the earth trying to get back the original files. There's a point where something has been over-compressed so much, the material has been stomped-on into oblivion. No amount of processing can really help. This is a little bit like "trying to get a cow back out of hamburger."

Mm, I think I did? -- "If often have to deal with already rendered and highly compressed video files". It means 720p videos that have no other (and better) copies, 1080p videos that have made a couple of trips in the wrong hands, compressed again, etc.

I was looking for a codec type, which is a key factor in workflow. What you describe is a hopeless situation. If you have 72P H.264 material that's been compressed twice, there's nothing you can do. I would give up and live with it, or I'd go to the ends of the earth trying to get back the original files. There's a point where something has been over-compressed so much, the material has been stomped-on into oblivion. No amount of processing can really help. This is a little bit like "trying to get a cow back out of hamburger."

Yeah I know. But that's the extreme. Generally, even the 720p ones are relatively clean, and not having the artifacts stretched looks way better.

I tested a couple of things with the Media Management I learned about recently because of this thread. I didn't upscaled my 1080p to 4k (but did it for the 720p ones), just denoised it and removed the compression artifacts with the external tool, then used the super scale option in Resolve. It's not as good as what I was doing, but WAY better than just dropping these and have big blocks of the edges stretched, and it takes way less time.