Ingest, proxies and the conform
This service starts with footage that already exists. Every other job on this site begins with a shoot. This one begins with a drive, and the first hour is not creative: verified copies onto two volumes plus one off site, checksums compared rather than file counts eyeballed, and bins named to a scheme that still reads in a year.
A proxy is a small, cheap-to-decode copy of every clip, so an editor cuts at full speed on material a workstation cannot play back in real time. The cut is built against proxies, which is the offline. Once the edit is signed off the timeline relinks to the full-resolution originals, which is the conform. Adding a scene after that means re-running colour and graphics on everything downstream.
The MGK Lost Americana behind the scenes work is a post-first job of this shape. The material existed. Every decision left was about what it became.
Codecs, and why the master file is not the upload
Acquisition, mezzanine and delivery codecs do three different jobs. Mezzanine codecs such as ProRes and DNxHR store every frame independently, which makes files heavy and easy to cut, re-grade and hand between facilities. Delivery codecs such as H.264, H.265 and AV1 describe most frames as differences from their neighbours, which makes files small enough to stream and unpleasant to re-encode.
So one graded master file is held at mezzanine quality and every published version is derived from it. Re-encoding a delivery file costs quality every time. Subsampling lands at the same point: 4:2:0 keeps full luminance detail and discards most colour detail, which is fine for playback and painful under a heavy grade or a key.
YouTube publishes what it wants. MP4 with the moov atom at the front and no edit lists, H.264 High Profile, progressive scan, closed GOP at half the frame rate, two consecutive B frames, CABAC, 4:2:0, AAC-LC audio at 48 kHz. Recommended SDR bitrates at 24 to 30 fps are 8 Mbps for 1080p, 16 Mbps for 1440p and 35 to 45 Mbps for 4K.
Log footage, LUTs and the grade
Colour is two operations. Balancing comes first: matching shots inside a scene so skin tone, white point and exposure hold across cuts made from three cameras and four hours of moving daylight. Grading is the look applied on top, and a look applied before the balance spreads the inconsistency rather than hiding it.
Log footage looks flat because it is supposed to. A log curve stores more highlight and shadow information at the cost of on-screen contrast, and it is graded, not watched. A LUT is a lookup table mapping one set of colour values onto another, so it is a starting point, and a viewing LUT on a monitor is not the grade that ships.
Standard high definition colour is defined by ITU-R BT.709, which YouTube also names as its recommended colour space for SDR uploads. High dynamic range works to ITU-R BT.2100, which defines the HLG and PQ transfer functions. An HDR grade does not remove the need for an SDR version.
The mix, and loudness measured to a standard
Dialogue is cleaned before anything else happens to it. Noise reduction, de-essing, level matching between takes and between microphones, then room tone under the joins so the silence does not jump at every cut. Sound design and music edited to picture follow, then a mix that holds the voice above the bed wherever somebody speaks.
Loudness is measured, not judged by ear. ITU-R BS.1770 defines the algorithms for measuring programme loudness and true-peak level. Targets differ by destination, which is the part clients are most often surprised by, so the target is set per deliverable before the mix starts rather than argued about after.
Ask for stems. Separate dialogue, music and effects tracks cost nothing to export while the session is open and save a full re-mix when a version needs a different voiceover or a language swap.
Graphics, titling and the caption file that ships with the master
Title and lower-third systems are built once and reused. Type, weight, colour, margins and timing get set as a template, so the twelfth video matches the first and a misspelled name is corrected in ten seconds. Logo and data animation and clean-up work such as screen replacement sit in the same stage.
Captions here are a file format question. A sidecar file, usually SRT or WebVTT, travels alongside the video and can be corrected or translated after delivery without touching the picture. Embedded broadcast caption data rides inside the file for destinations that require it. Open captions are burned into the picture and cannot be removed or translated afterwards. Captions as a sound-off comprehension device is a different question from which file a platform accepts, and accessibility policy for internal distribution is a third.
What counts as one round
A revision round is one consolidated set of notes, from one approver, addressed together. Notes arriving one at a time across four days are not a round. They are four rounds, and they are the most common reason a schedule slips without anyone deciding to slip it. Collect notes on a review platform with frame-accurate comments, so a note reads 00:42, cut two frames earlier.
Rounds are counted and scoped before the edit begins. Changes inside the existing structure are revisions. A new voiceover script, a different music track after the mix, or a scene that was never in the assembly are scope, and they get quoted rather than absorbed. Numbered versions with dates, and nothing called final, stop anyone approving v3 while the edit has moved to v5.
For artists, managers and labels the schedule runs backwards from a release date rather than forwards from a brief, and the date is usually locked before the drive arrives, so the person at the label who can approve is named on day one.
Archive is a decision, not a default
Storage is not archive. A drive on a shelf is a single point of failure, and an unindexed folder is not retrievable at any size. Three things get decided at the start of the job: how long camera originals are kept, how long project files are kept, and who holds them.
Working storage stays online while the job is live and for a defined window after delivery, which is when late changes actually turn up. After that, camera originals either move to cold storage, meaning LTO tape or an archive tier priced for infrequent reads, or they go back to the client on media the client owns.
Handing over files you can actually reuse
The deliverable is a matrix, not a file. One graded master at mezzanine quality sits at the top. Under it: per-destination encodes derived from that master, a caption sidecar wherever the destination takes one, textless versions of anything carrying burned-in type, alternate aspect ratios cut rather than cropped, and stems where a future version is likely.
Brand templates and editable project files belong on that list whenever the client will maintain the work in house. A supplied brand kit means type and colour arrive as a specification instead of a guess, and a returned project file means a title can be changed next quarter without the job rejoining a queue. Agree the application, the font licences and whether linked media travels with it, because a project file without its media is a list of instructions.
Post runs remotely, so this is the one service here with no travel line on it. Files from Simi Valley or Oxnard are handled the same way as files sent from Burbank or another state. Studio 47 finishes work shot by other teams across Ventura County and Los Angeles County, runs the job from the production side when the shoot is ours, and the service list shows what feeds into post.
Frequently asked questions
How much does video editing cost?
Footage volume sets the floor, because somebody watches all of it before a cut exists. On top of that sit assembly complexity, how much motion graphics work the piece carries, whether colour goes past balancing, how much dialogue cleanup the sound needs, and how many revision rounds are scoped. Two pricing models exist, per finished minute and hourly, and each rewards different work.
How long does video editing take?
Published ratios put roughly an hour of edit behind each finished minute for a simple cut, one to five hours a minute for corporate work, and considerably more once effects, motion graphics or a heavy grade are involved. Review adds calendar rather than hours. Adding a scene after the conform re-runs colour and graphics on everything downstream.
Can you edit footage we shot ourselves?
Yes, and this service starts there. Send the original camera files rather than compressed exports, any separately recorded audio, the brand assets you want used, and one line naming the destination and the deadline. Studio 47 verifies copies onto two volumes plus one off site before anything is cut. Mixed formats and frame rates get resolved at ingest by settling on one timeline standard.
What is the difference between video editing and post-production?
Editing is the cut. Post-production is everything that turns the cut into a finished file: colour, sound design and mix, motion graphics and titling, captions, then the encodes each destination asks for. The cut is built against proxies, which is the offline, and relinks to the full-resolution originals afterwards, which is the conform. Studio 47 scopes those stages separately, since a job can stop at the cut.
What is colour grading and does my video need it?
Two operations sit behind the word. Balancing matches shots inside a scene so skin tone, white point and exposure hold across cuts made from three cameras and four hours of moving daylight. Grading is the look applied on top. Any job cut from more than one camera needs the first, because a look applied to unbalanced material spreads the mismatch rather than hiding it.
How many revision rounds come with an edit?
A fixed number, agreed before the edit begins rather than after. The definition matters more than the count. One round is one consolidated set of timestamped notes from one named approver, addressed together. Notes trickling in one at a time across four days are four rounds, and that is what breaks a locked release date more often than the work does.
What file formats and versions do you deliver?
A matrix rather than a file. One graded master at mezzanine quality sits at the top and every published version derives from it, because re-encoding a delivery file costs quality each time. Under it: per-destination encodes, a caption sidecar such as SRT or WebVTT where the destination takes one, textless versions of anything carrying burned-in type, alternate ratios cut rather than cropped for social, and audio stems.