Why Is Your Recording File So Huge? Bitrate, Rate Control and Resolution / Frame Rate Diagnosis Guide

Key takeaway: A recording file that is several gigabytes is not necessarily broken: file size is bitrate multiplied by time, so the first job is arithmetic, not guessing. Compare megabytes per minute against the bitrate the settings should produce, then work through rate-control mode, the CQP or CRF value, resolution and frame rate, and the encoder, changing one setting at a time.
A recording comes out several gigabytes in size. Ten minutes of gameplay fills more than three gigabytes. An hour of footage fills tens of gigabytes, and the disk fills up faster than expected. The recording itself plays back perfectly: the picture is normal, the sound is normal, the length is right, and nothing in the file looks damaged. That combination — a file that feels unreasonably large and plays back without any fault — is the whole problem this guide addresses, and the combination itself is the first piece of evidence.
The instinct in this situation is to blame the recording software, or the graphics card, or the disk. The software is reinstalled because it is the part the reader can see, the card is suspected because it does the encoding, or a bigger disk is bought because the disk is the part that ran out of room. All three instincts are premature. A file that plays back normally is not a broken file. Its size is the arithmetic result of the settings that produced it: the bitrate, the rate-control mode, the resolution, the frame rate, and the encoder. Any one of those settings can be set high enough to produce a very large file while the recording itself is completely healthy. The first job is therefore not to replace anything, but to calculate what size the settings should have produced, and to compare that number with the size of the file on the disk.
This guide covers only a narrow shape. Four conditions must all be present. First, the file is a recording made by software such as OBS, ShadowPlay, or Game Bar, and it is unreasonably large: a short clip of several gigabytes, or an hour of footage running into tens of gigabytes, with the disk filling up quickly. Second, the file plays back normally and its content is complete: the picture is normal, the sound is normal, and the length is correct. Third, the problem is size, not stuttering or silence: recording that stutters or drops frames during capture belongs to the encoding-overload layer, a recording file with no sound belongs to the silent-recording family, and a picture that freezes while the sound continues belongs to the mirroring layer. Fourth, the reader has settings to compare against: the rate-control mode in use, the bitrate or CQ value, the resolution, and the frame rate. If none of those settings are known, the first step is to write down the baseline before anything else is changed.
File size is arithmetic, not luck
File size is not random. It follows one formula: file size in megabytes is approximately the bitrate in megabits per second, multiplied by the number of seconds, divided by eight. A guide published by castr uses the same formula to build a reference table: recording at 1080p60 with 30,000 to 50,000 Kbps produces about 14 to 23 gigabytes per hour, and recording at 4K60 with 80,000 to 130,000 Kbps produces about 36 to 59 gigabytes per hour. The arithmetic can be checked directly: 50,000 Kbps for ten minutes is 50,000 divided by eight, multiplied by 600 seconds, divided by 1,024, which comes to about 3.6 gigabytes (3,662 MB). That works out to about 366 megabytes per minute, against the 375 megabytes per minute the 50 Mbps setting should produce: the two figures are close enough once rounding is allowed for. That number is not a fault. It is the arithmetic result of the settings themselves. The first step is always to calculate, not to guess.
That reference table needs a clear label, and this guide gives it one. The castr table is a guide's starting figures, not a measured standard and not a universal table of normal file sizes. No outside source consulted for this guide provides a universal table of what a normal recording file size should be, and this guide does not invent one. The standard in this guide is always the reader's own arithmetic comparison: the file's megabytes per minute set against the megabytes per minute the reader's own settings should produce. A file that matches its own settings is large because the settings are large, and a file that does not match its own settings points somewhere else. No fixed table can replace that comparison, because two recordings with the same resolution and frame rate can legitimately differ in size when their rate-control mode, their quality value, and the complexity of the scene are different.
There is a forum case that makes the same point from the other direction. On the OBS official forum, a user asked whether a file of about 12 gigabytes for 75 to 80 minutes of 1080p30 recording was too large. A forum moderator answered in plain words: "There is nothing 'wrong' with your results." The file size matched the settings, so there was no fault to find. That is a forum case, not a conclusion from this site, and it is used here only to show that large does not mean broken. It must not be written as this site's standard answer, and it must not be turned into a rule that any particular file size is normal. The only standard this guide uses is the reader's own calculation.
CBR was designed for streaming, not for recording
The single most important sentence in this guide is that using the wrong rate-control mode makes a large file expected, not abnormal. CBR, constant bitrate, was designed for live streaming, because a streaming server needs a steady and predictable bitrate. A moderator on the OBS official forum states this directly: CBR is the right choice for streaming, and recording should use CQP or CRF, which are quality-target modes. CBR does not care whether the scene is simple or complex. It uses the same bitrate at every moment: it wastes space on a simple scene and gives too little space to a complex scene. Recording with CBR at a high bitrate produces a large file by design, not by accident.
That distinction explains why a recording made with CBR can be large even when nothing went wrong. The mode itself is the cause. A simple scene such as a static menu or a slow puzzle game needs very little data to look good, but CBR keeps spending the full bitrate anyway, because the mode has no way to spend less. A complex scene such as a fast action game with a lot of movement needs more data, but CBR cannot give it more than the fixed number. The file ends up large in the easy parts and starved in the hard parts, which is the worst of both directions. Switching the mode is therefore not a small tweak. It changes the whole logic of how the encoder decides what to spend.
The practical instruction follows from the mechanism. In OBS, the rate-control mode is a setting in the output section, and it must be read before it is changed. If the mode is CBR, the diagnosis has already found its first suspect, and the comparison is to record a short test clip in CQP or CRF with the same scene and compare the megabytes per minute. If the mode is already CQP or CRF, the diagnosis moves on to the quality value, because the mode alone is not the whole story.
CQP and CRF: fixed quality, variable bitrate
CQP and CRF are quality-target modes. CQP is used with hardware encoders such as NVENC, and CRF is used with the x264 software encoder. The logic is the opposite of CBR: the quality stays fixed, and the bitrate follows the scene. A simple scene uses less data, a complex scene uses more data, and the picture quality stays the same. That is why the same recording can come out much smaller in CQP or CRF than in CBR without looking worse: the easy parts stop wasting space.
The numbers run in a direction that must not be written backwards. A smaller number means higher quality and a larger file. A larger number means lower quality and a smaller file. The OBS forum states that CQP 22 is a good starting point, that 16 is close to visually lossless but produces a much larger file, and that 12 and below is only suitable for a master copy that will be edited and compressed again later. The salivity guide adds that a CQ or CRF value of 20 to 23 is a balance point, that 24 to 25 will shrink the file noticeably while the picture quality drops only slightly, and that values below 16 are not recommended for ordinary recording. The castr guide gives a recommended range of CQP 18 to 23. These are guide figures from different sources, and they agree on the direction even when their exact numbers differ slightly. The reader should treat them as starting points for a comparison, not as a specification.
One piece of forum experience needs its source label kept attached to it at all times. Forum users report that raising the CQ value by 3 to 4 roughly halves the file size. That is a forum user's experience, stated on the OBS forum, not a measurement made by this site, and this guide reports it only with that label. It must never be written as this site's conclusion. The reason for the caution is that the same forum thread also reports a case where CQP 25 produced a bitrate of 60k to 70k while looking worse than CBR at 40k. The two reports do not contradict each other: there is no fixed conversion between a CQ value and an actual bitrate, because the actual bitrate depends on how complex the scene is. A CQ value cannot be written as corresponding to a fixed bitrate, and a change of 3 to 4 cannot be promised to halve any particular file. The only honest way to find out what a change does is to record a short test clip in the same scene and compare the megabytes per minute.
Lossless: mathematically lossless, enormously large
OBS simple mode offers three quality options, and the techbloat guide quotes them in their own words: High Quality, Medium File Size, which is the reasonable default; Indistinguishable Quality, Large File Size; and Lossless Quality, Tremendously Large File Size. The wording is the point. Lossless is not the best setting. It is the largest setting. The castr guide states the same thing from the other side: lossless is mathematically lossless compression, the file is enormously large, and it must never be used for streaming.
Lossless means that no picture information is thrown away at all. Every pixel is kept exactly. That is useful for one purpose only: a master copy that will be edited and compressed again later, where the editor wants to start from the fullest possible source. For ordinary recording, lossless is the wrong tool, because it produces a file many times larger than a high-quality CQP or CRF recording that looks the same to the eye. The difference between lossless and a CQ value of 16 is small on screen and large on disk. A reader who finds Lossless selected in the settings has found the answer: the file is large because the setting asks for everything to be kept, and the remedy is to change the setting, not to buy a bigger disk.
The check is simple and decisive. Open the output settings and read the quality option or the rate-control mode. If it says Lossless, the file size is explained, and the comparison is to record a short test clip in CQP at 22 with the same scene and compare the megabytes per minute. If the mode is not Lossless, the diagnosis moves on. This step only reads the mode. It does not change anything else at the same time, or the result cannot be attributed to the change.
Resolution and frame rate: more pixels, more data
Resolution and frame rate decide directly how much data must be written. A 4K frame has four times as many pixels as a 1080p frame, so it carries roughly four times as much picture information before compression. A 60 frames per second recording has twice as many frames as a 30 frames per second recording, so it carries roughly twice as much movement information. Raising either one raises the amount of data the encoder must handle, and the file grows with it. The OBS forum thread states this plainly: bitrate is not independent of resolution and frame rate, and raising the resolution or frame rate requires raising the bitrate to maintain the same picture quality.
The castr reference table shows the same relationship in numbers, and the numbers are guide figures, not a standard. Recording at 720p30 with 8,000 to 12,000 Kbps produces about 4 to 5 gigabytes per hour. Recording at 1080p30 with 20,000 to 30,000 Kbps produces about 9 to 14 gigabytes per hour. Recording at 1080p60 with 30,000 to 50,000 Kbps produces about 14 to 23 gigabytes per hour. Recording at 1440p60 with 40,000 to 60,000 Kbps produces about 18 to 27 gigabytes per hour. Recording at 4K60 with 80,000 to 130,000 Kbps produces about 36 to 59 gigabytes per hour. Each step up in resolution or frame rate moves the whole range upward. Those figures are a guide's starting points for choosing settings, and they must not be quoted as a universal table of normal file sizes. A reader whose recording is more complex than the guide's example scene will legitimately produce a larger file at the same settings, and a reader whose recording is simpler will legitimately produce a smaller one.
Dynamic resolution adds a further wrinkle. Some games and some recording setups change the rendering resolution while the recording is running, which makes the file size vary from one part of the recording to another. A file that is large overall may have been recorded at a higher resolution for part of its length than the settings page suggests. The comparison that catches this is the short test clip: record one minute of the same scene at a fixed resolution and frame rate, and compare its megabytes per minute with the full file's average. If the test clip is much smaller per minute than the full file, the full file's size came from something that changed during the recording, and the resolution and frame rate layer is where the difference sits.
Audio is not the cause of a large file, and the arithmetic shows why. An audio track runs at 160 to 320 kilobits per second, while the video track runs into tens of thousands of kilobits per second. Against a video track of tens of thousands of kilobits per second, the audio track is roughly one hundredth of the total. A file that is several gigabytes large cannot be explained by its audio track, and changing the audio bitrate will not shrink it in any meaningful way. Readers who suspect the audio track should run the arithmetic first: the audio track's share is too small to matter, and the video settings are where the size lives.
The five suspect layers
| Layer | What it means | Decisive comparison |
|---|---|---|
| Arithmetic layer | The file size does or does not match the bitrate the settings should produce: a match means the settings themselves are that large, a mismatch points at the rate-control or lossless layer | Megabytes per minute arithmetic comparison: file megabytes divided by minutes, set against the setting's megabits per second multiplied by 60 and divided by 8 |
| Rate-control mode layer | CBR, CQP, CRF, and Lossless produce very different file sizes: CBR is a streaming mode, Lossless is a master-copy mode | Rate-control mode comparison: read which mode the settings use |
| Setting-value layer | The bitrate number itself is too high, or the CQ or CRF value is too low, with values below 16 close to visually lossless | CQ or CRF value comparison, plus bitrate number comparison |
| Resolution and frame rate layer | 4K and 60 frames per second need much more data by themselves, and dynamic resolution makes the file size float | Resolution and frame rate comparison: drop one level, record a short test clip, and see how far the megabytes per minute fall |
| Encoder layer | H.264 versus HEVC, and the encoder preset, affect the file size | Encoder comparison: the guide figure says HEVC saves roughly three to five tenths, which is a guide figure, not a measurement |
Each layer has its own comparison, and no single comparison covers all five. The arithmetic comparison separates a file that matches its settings from a file that does not. The mode comparison separates a streaming mode from a recording mode. The value comparison separates a quality number that is too low from one that is reasonable. The resolution comparison separates a file that is large because the picture is large from a file that is large because the encoding is wasteful. The encoder comparison separates the codec from everything else. The order in the seven steps below is the order that wastes the least time.
Seven diagnostic steps
Step 1 — Define the shape
Confirm the four shape conditions: the recording file is unreasonably large, it plays back normally with its content complete, the problem is size rather than stuttering or silence or a file that will not open, and the reader has settings to compare against. Write down the baseline before changing anything: the file size in megabytes or gigabytes, the length in minutes, the recording software, the rate-control mode (CBR, CQP, CRF, or Lossless), the bitrate or CQ value, the resolution, the frame rate, and the encoder (H.264 or HEVC). A diagnosis built on a vague description such as "the file is too big" cannot be checked later, because it does not say what the settings were or how large the file actually was. The baseline is the comparison material for every step that follows.
Step 2 — Compare megabytes per minute against the settings
Divide the file size in megabytes by the length in minutes to get the megabytes per minute. Then calculate what the settings should produce: the bitrate in megabits per second, multiplied by 60, divided by 8. If the two numbers agree, the file is large because the settings are large, and the problem is in the settings, not in the file. If the two numbers differ widely, the fault points at the rate-control or lossless layer, and the diagnosis moves to Step 3. This step is the first key comparison in the whole guide: calculate first, and do not call a file unreasonably large on feeling alone. A file of 12 gigabytes for 75 to 80 minutes of 1080p30 recording works out to 154–164 MB/min, and the forum moderator's answer was that nothing was wrong with that result, because the number matched the settings. The reader's own numbers are the only standard that matters.
Step 3 — Compare the rate-control mode
Read the settings to see which mode is in use. If the mode is Lossless, the answer is here: a lossless file is enormously large by design, and it is not a fault. If the mode is CBR at a high bitrate, the answer is also here: CBR is a streaming mode, and recording should not use it. If the mode is CQP or CRF, the diagnosis moves to Step 4. This step only reads the mode. Do not change other settings at the same time, or the result cannot be attributed to the change that produced it. Record a short test clip in the same scene after changing the mode, and compare the megabytes per minute with the baseline. A large drop points at the mode layer. A small drop means the mode was not the main cause, and the diagnosis moves on.
Step 4 — Compare the CQ or CRF value
A CQP or CRF value below 16 is close to visually lossless, and a large file is expected at that value. A value of 22 is the starting point named on the OBS forum. A value of 24 to 25 will shrink the file noticeably, while the picture quality drops only slightly, according to the salivity guide. When changing the value, change it by 3 to 4 at a time — the forum experience that raising the CQ value by 3 to 4 roughly halves the file size is a forum user's report, not a measurement by this site — and record a short test clip in the same scene to compare. Do not judge the change by feeling. Compare the megabytes per minute before and after, and keep the clip that looks acceptable at the smallest size. Note that the CQ value and the actual bitrate have no fixed conversion: the same CQ value produces a different bitrate in a simple scene and in a complex scene, so the test clip must use the same scene as the baseline, or the comparison means nothing.
Step 5 — Compare the resolution and frame rate
Drop one level and record a short test clip: 4K down to 1440p or 1080p, or 60 frames per second down to 30 frames per second, and see how far the megabytes per minute fall. A large drop points at the resolution and frame rate layer: the picture itself needs that much data. A small drop means the problem sits in the rate-control or encoder layer, and the diagnosis moves to Step 6. Change only one thing at a time. Do not drop the resolution and the frame rate and the quality value all at once, or the result cannot be attributed to any one of them. The test clip must use the same scene as the baseline, because a quiet scene and a busy scene produce very different file sizes at the same settings.
Step 6 — Compare the encoder
Switch from H.264 to HEVC and record a short test clip in the same scene to compare. The castr guide says HEVC saves roughly three to five tenths compared with H.264, and that figure is a guide figure, not a measurement standard, so it must be reported with its source label. HEVC also carries two cautions that must be checked at the same time: the encoding load is heavier, so a weaker machine may struggle, and not every editing program accepts HEVC files, so the file may be smaller but harder to edit. On the encoder preset side, the OBS forum warns against using Max Quality and Psychovisual Tuning, because those two options add encoding load. That warning belongs to the encoding-overload layer, not to this guide, and it is not expanded here. If the HEVC test clip is much smaller per minute than the H.264 baseline, the encoder layer is where the saving sits. If it is barely smaller, the size is set by the rate control and the resolution, and the encoder is not the main cause.
Step 7 — Stop if the fault is still undivided
If the seven steps are complete, the arithmetic agrees, and the settings have been confirmed, then a large file is the settings' own result, not a fault. To save space, change the settings and record again; do not delete the wrong files. An old file that needs to be smaller must be recompressed, which is a separate job, not a repair. Before changing any setting, back up the original settings, and write down which step produced which number, because those numbers are the comparison material. If the fault still cannot be divided, or if a hardware fault is suspected, stop and bring the written record to Room 1209, Dali Plaza, Shek Mun for inspection. The fee for inspection is given in the closing section below.
Boundary cases
A recording that crashes or loses power part-way through, leaving a file that will not open, is an index and truncation problem, not a file-size problem. The file structure is damaged, and the file size in that guide is only a diagnostic field. A recording or stream that stutters, drops frames, or shows an encoder-overload warning is an encoding-overload problem: dropped frames sit at the load layer, and the file size is not the main subject. A background-buffer recording that stutters is a background-buffer problem, not a file-size problem. A recording file that is silent all the way through is a silent-recording problem, at the audio-track layer, not the size layer. A recording file that goes silent part-way through is a timing-shape problem, not a size problem. An editing export that fails or stops part-way through is an export-failure problem, not a recording-file problem. An exported finished file with broken pictures is a broken-output problem, not a recording-file problem. A player that plays smoothly while the editing program stutters is a variable-frame-rate problem, at the timestamp layer, not the size layer. A file that opens but has silent passages in some parts is a track or import problem, not a size problem. Each of these has its own guide, and the file-size diagnosis in this guide does not apply to them.
Six common misconceptions
- "A large file means the recording is broken." Wrong. File size is the arithmetic result of bitrate multiplied by time. When the calculation agrees with the settings, the file is large because the settings are large, not because anything broke. The OBS forum case of about 12 gigabytes for 75 to 80 minutes of 1080p30 recording ended with the moderator stating that nothing was wrong with the result.
- "CBR is the most stable choice for recording." Wrong. CBR was designed for a streaming server that needs a steady and predictable bitrate. Recording should use CQP or CRF, which spend less data on a simple scene and more on a complex scene while keeping the quality fixed.
- "Lossless is always the best." Wrong. Lossless produces an enormously large file and is only suitable for a master copy that will be edited and compressed again later. For ordinary recording, lossless fills the disk quickly. OBS simple mode labels it Tremendously Large File Size in its own words.
- "Lowering the resolution always halves the file size." Wrong. How much is saved must be calculated: the resolution, the frame rate, and the complexity of the scene all affect the result together. Drop one level, record a short test clip, and compare the megabytes per minute before writing any figure for the saving.
- "Switching to HEVC immediately halves the file size." Wrong. The guide figure is roughly three to five tenths, and it is an approximate figure from a guide, not a measurement standard. Encoder support and editing-program support must also be checked, because not every machine and not every program works well with HEVC.
- "A large file means I need to buy a bigger hard drive." Wrong. Buying a disk is capacity management, not diagnosis. Calculate first, then change the settings, and the file size may already have fallen, without buying anything.
What is confirmed, what is only a guide figure
The confirmed parts of this guide are the ones the outside sources state in their own words: the file-size formula of bitrate multiplied by time divided by eight, the castr reference table as a guide's starting figures, the OBS forum's statement that CBR belongs to streaming and recording should use CQP or CRF, the forum's starting point of CQP 22 with 16 close to visually lossless and 12 and below for master copies only, the techbloat quotation of the three OBS simple-mode quality options, and the forum case in which a moderator stated that about 12 gigabytes for 75 to 80 minutes of 1080p30 recording had nothing wrong with it.
Three figures must be labelled openly as guide figures or forum experience, and this guide labels them here rather than presenting them as measurements. The castr reference table — 1080p60 at 14 to 23 gigabytes per hour, 4K60 at 36 to 59 gigabytes per hour, and the other rows — is a guide's starting figures, not a universal standard, and no universal table of normal file sizes exists in the sources consulted. The forum experience that raising the CQ value by 3 to 4 roughly halves the file size is a forum user's report, not a conclusion from this site. The HEVC saving of roughly three to five tenths is a castr guide figure, not a measurement result. No figure in milliseconds, no ranking, no case count, no success rate, and no fee beyond the site's published three-tier fee appears in this guide, because none is supported by the sources used. The only arithmetic this guide uses is the reader's own megabytes-per-minute comparison.
Sources and what they do and do not establish
The OBS official forum thread "File size way to big" states that a user recording 1080p30 for 75 to 80 minutes produced a file of about 12 gigabytes, and that a moderator answered that there was nothing wrong with the result, because the file size matched the settings; that CBR is for streaming and recording should use CQP or CRF; that CQP 22 is a good starting point, that 16 is close to visually lossless but produces a much larger file, and that 12 is only suitable for a master copy that will be edited and compressed again; and that Max Quality and Psychovisual Tuning should not be used, because they add encoding load, which belongs to the encoding-overload layer. It is used here for the rate-control mode, the quality values, and the forum case.
The OBS official forum thread "Isn't it too much of a size?" states that CQP is a quality-target mode, using less data on a simple scene and more on a complex scene; that raising the CQ value by 3 to 4 roughly halves the file size, as a forum user's experience; that at CQP 25 the bitrate can reach 60k to 70k while looking worse than CBR at 40k, which shows that a CQ value must not be pushed up blindly; and that bitrate is not independent of resolution and frame rate, so raising the resolution or frame rate requires raising the bitrate to maintain the same picture quality. It is used here for the CQ-value behaviour and the warning against blindly raising the value.
The castr guide "Best OBS Settings for Streaming and Recording" states that recording should use CQP with a GPU encoder or CRF with x264, not CBR; that CQP 18 to 23 is the recommended range; that the reference table runs from 720p30 at 8,000 to 12,000 Kbps and about 4 to 5 gigabytes per hour, through 1080p30 at 20,000 to 30,000 Kbps and about 9 to 14 gigabytes per hour, 1080p60 at 30,000 to 50,000 Kbps and about 14 to 23 gigabytes per hour, and 1440p60 at 40,000 to 60,000 Kbps and about 18 to 27 gigabytes per hour, to 4K60 at 80,000 to 130,000 Kbps and about 36 to 59 gigabytes per hour; that HEVC saves roughly three to five tenths compared with H.264; and that lossless files are enormously large and must never be used for streaming. It is used here for the reference table and the HEVC figure, both labelled as guide figures.
The salivity guide "How to Reduce OBS Studio Recording File Size" states that the main cause of a large file is the bitrate setting, whether lossless or an excessively high CBR; that the rate control should be changed to CQP or CRF; and that a CQ or CRF value of 20 to 23 is a balance point, that 24 to 25 will shrink the file noticeably while the picture quality drops only slightly, and that values below 16 are not recommended. It is used here for the value ranges.
The techbloat guide "OBS Studio How to Change Bitrate" states the three OBS simple-mode quality options in their own words — High Quality, Medium File Size; Indistinguishable Quality, Large File Size; and Lossless Quality, Tremendously Large File Size — that the recording bitrate and the streaming bitrate are separate settings in advanced mode, and that CQP applies to hardware encoders such as NVENC while CRF applies to x264. It is used here for the quality options and the encoder split.
There is a conflict between the sources that this guide states rather than resolves. One forum thread calls CQP 22 a good starting point, while another reports that CQP 25 can produce a bitrate of 60k to 70k while looking worse than CBR at 40k. The two reports do not contradict each other: there is no fixed conversion between a CQ value and an actual bitrate, because the actual bitrate depends on how complex the scene is. A CQ value must never be written as corresponding to a fixed bitrate. The castr reference table and the forum case figure of about 12 gigabytes for 75 to 80 minutes are two groups of numbers from two different sources, and they must not be mixed together to calculate an average. The standard in this guide is always the reader's own arithmetic comparison.
FAQ
Does a huge recording file mean the recording is broken?
No. File size is the arithmetic result of bitrate multiplied by time. When the calculation agrees with the settings, the file is large because the settings are large, not because anything broke. On the OBS forum, a user asked about a file of about 12 gigabytes for 75 to 80 minutes of 1080p30 recording, and a moderator answered that there was nothing wrong with the result, because the file size matched the settings. A large file that plays back normally is a settings question first, and a fault question only after the arithmetic disagrees.
Is CBR the most stable choice for recording?
No. CBR was designed for live streaming, because a streaming server needs a steady and predictable bitrate. Recording should use CQP or CRF, which are quality-target modes: a simple scene uses less data, a complex scene uses more data, and the picture quality stays fixed. CBR spends the same bitrate on every scene, wasting space on simple scenes and giving too little to complex ones. A large file recorded with CBR at a high bitrate is expected, not abnormal.
Is lossless always the best quality setting?
No. Lossless produces an enormously large file, and it is only suitable for a master copy that will be edited and compressed again later. OBS simple mode labels it Lossless Quality, Tremendously Large File Size, in its own words. For ordinary recording, a high-quality CQP or CRF recording looks the same to the eye at a fraction of the size. Lossless is the largest setting, not the best setting.
Will lowering the resolution always halve the file size?
No. How much is saved must be calculated, because the resolution, the frame rate, and the complexity of the scene all affect the result together. Drop one level, record a short test clip in the same scene, and compare the megabytes per minute before and after. Only that comparison shows the actual saving for that recording. No fixed figure for the saving can be promised in advance.
Will switching to HEVC immediately halve the file size?
No. The guide figure is roughly three to five tenths saved compared with H.264, and that figure comes from the castr guide, not from a measurement. It is an approximate figure, and the actual saving depends on the scene, the settings, and the encoder. HEVC also needs heavier encoding work and is not accepted by every editing program, so the encoder and the editing workflow must be checked at the same time as the file size.
Does a huge recording file mean I need to buy a bigger hard drive?
No. Buying a disk is capacity management, not diagnosis. Calculate first, then change the settings, and the file size may already have fallen. Work through the seven steps: compare megabytes per minute against the settings, check the rate-control mode, check the CQ or CRF value, drop the resolution or frame rate one level, and compare the encoder. Only after the settings have been checked and the file is still unreasonably large is it worth considering more storage, and even then the old files still need to be recompressed separately if space is the goal.
If the fault still cannot be divided
Bring the written record described in Step 7 to Room 1209, Dali Plaza, Shek Mun for inspection. The telephone number is 2186 8010 and WhatsApp is 6558 6806. There is no Mong Kok shop, and there is no on-site service. The site's published three-tier fee is HK$400–780, HK$480–1,380, and HK$680–2,800, an overall range of HK$400–2,800, with the actual fee quoted after inspection. Do not delete the original settings; they are the comparison material.