Recherche avancée

Sur d’autres sites (6987)

• Linear Attribution Model : What Is It and How Does It Work ?

  16 février 2024, par Erin

  Want a more in-depth way to understand the effectiveness of your marketing campaigns ? Then, the linear attribution model could be the answer.

  Although you can choose from several different attribution models, a linear model is ideal for giving value to every touchpoint along the customer journey. It can help you identify your most effective marketing channels and optimise your campaigns. 

  So, without further ado, let’s explore what a linear attribution model is, when you should use it and how you can get started. 

  What is a linear attribution model ?

  A linear attribution model is a multi-touch method of marketing attribution where equal credit is given to each touchpoint. Every marketing channel used across the entire customer journey gets credit, and each is considered equally important. 

  So, if a potential customer has four interactions before converting, each channel gets 25% of the credit.

  

  Let’s look at how linear attribution works in practice using a hypothetical example of a marketing manager, Sally, who is looking for an alternative to Google Analytics. 

  Sally starts her conversion path by reading a Matomo article comparing Matomo to Google Analytics she finds when searching on Google. A few days later she signs up for a webinar she saw on Matomo’s LinkedIn page. Two weeks later, Sally gets a sign-off from her boss and decides to go ahead with Matomo. She visits the website and starts a free trial by clicking on one of the paid Google Ads. 

  Using a linear attribution model, we credit each of the channels Sally uses (organic traffic, organic social, and paid ads), ensuring no channel is overlooked in our marketing analysis. 

  Are there other types of attribution models ?

  Absolutely. There are several common types of attribution models marketing managers can use to measure the impact of channels in different ways. 

  

  • First interaction : Also called a first-touch attribution model, this method gives all the credit to the first channel in the customer journey. This model is great for optimising the top of your sales funnel.
  • Last interaction : Also called a last-touch attribution model, this approach gives all the credit to the last channel the customer interacts with. It’s a great model for optimising the bottom of your marketing funnel. 
  • Last non-direct interaction : This attribution model excludes direct traffic and credits the previous touchpoint. This is a fantastic alternative to a last-touch attribution model, especially if most customers visit your website before converting. 
  • Time decay attribution model : This model adjusts credit according to the order of the touchpoints. Those nearest the conversion get weighted the highest. 
  • Position-based attribution model : This model allocates 40% of the credit to the first and last touchpoints and splits the remaining 20% evenly between every other interaction.

  Why use a linear attribution model ?

  Marketing attribution is vital if you want to understand which parts of your marketing strategy are working. All of the attribution models described above can help you achieve this to some degree, but there are several reasons to choose a linear attribution model in particular. 

  It uses multi-touch attribution

  Unlike single-touch attribution models like first and last interaction, linear attribution is a multi-touch attribution model that considers every touchpoint. This is vital to get a complete picture of the modern customer journey, where customers interact with companies between 20 and 500 times. 

  Single-touch attribution models can be misleading by giving conversion credit to a single channel, especially if it was the customer’s last use. In our example above, Sally’s last interaction with our brand was through a paid ad, but it was hardly the most important. 

  It’s easy to understand

  Attribution models can be complicated, but linear attribution is easy to understand. Every touchpoint gets the same credit, allowing you to see how your entire marketing function works. This simplicity also makes it easy for marketers to take action. 

  It’s great for identifying effective marketing channels

  Because linear attribution is one of the few models that provides a complete view of the customer journey, it’s easy to identify your most common and influential touchpoints. 

  It accounts for the top and bottom of your funnel, so you can also categorise your marketing channels more effectively and make more informed decisions. For example, PPC ads may be a more common bottom-of-the-full touchpoint and should, therefore, not be used to target broad, top-of-funnel search terms.

  Are there any reasons not to use linear attribution ?

  Linear attribution isn’t perfect. Like all attribution models, it has its weaknesses. Specifically, linear attribution can be too simple, dilute conversion credit and unsuitable for long sales cycles.

  

  It can be too simple

  Linear attribution lacks nuance. It only considers touchpoints while ignoring other factors like brand image and your competitors. This is true for most attribution models, but it’s still important to point it out. 

  It can dilute conversion credit

  In reality, not every touchpoint impacts conversions to the same extent. In the example above, the social media post promoting the webinar may have been the most effective touchpoint, but we have no way of measuring this. 

  The risk with using a linear model is that credit can be underestimated and overestimated — especially if you have a long sales cycle. 

  It’s unsuitable for very long sales cycles

  Speaking of long sales cycles, linear attribution models won’t add much value if your customer journey contains dozens of different touchpoints. Credit will get diluted to the point where analysis becomes impossible, and the model will also struggle to measure the precise ways certain touchpoints impact conversions. 

  Should you use a linear attribution model ?

  A linear attribution model is a great choice for any company with shorter sales cycles or a reasonably straightforward customer journey that uses multiple marketing channels. In these cases, it helps you understand the contribution of each touchpoint and find your best channels. 

  It’s also a practical choice for small businesses and startups that don’t have a team of data scientists on staff or the budget to hire outside help. Because it’s so easy to set up and understand, anyone can start generating insights using this model. 

  How to set up a linear attribution model

  Are you sold on the idea of using a linear attribution model ? Then follow the steps below to get started :

  

  Choose a marketing attribution tool

  Given the market is worth $3.1 billion, you won’t be surprised to learn there are plenty of tools to choose from. But choose carefully. The tool you pick can significantly impact your success with attribution modelling. 

  Take Google Analytics, for instance. While GA4 offers several marketing attribution models for free, including linear attribution, it lacks accuracy due to cookie consent rejection and data sampling. 

  Accurate marketing attribution is included as a feature in Matomo Cloud and is available as a plugin for Matomo On-Premise users. We support a full range of attribution models that use 100% accurate data because we don’t use data sampling, and cookie consent isn’t an issue (with the exception of Germany and the UK). That means you can trust our insights.

  Matomo’s marketing attribution is available out of the box, and we also provide access to raw data, allowing you to develop your custom attribution model. 

  Collect data

  The quality of your marketing attribution also depends on the quality and quantity of your data. It’s why you need to avoid a platform that uses data sampling. 

  This should include :

  • General data from your analytics platform, like pages visited and forms filled
  • Goals and conversions, which we’ll discuss in more detail in the next step
  • Campaign tracking data so you can monitor the behaviour of traffic from different referral channels
  • Behavioural data from features like Heatmaps or Session Recordings

  Set up goals and conversions

  You can’t assign conversion values to customer journey touchpoints if you don’t have conversion goals in place. That’s why the next step of the process is to set up conversion tracking in your web analytics platform. 

  Depending on your type of business and the product you sell, conversions could take one of the following forms :

  • A product purchase
  • Signing up for a webinar
  • Downloading an ebook
  • Filling in a form
  • Starting a free trial

  Setting up these kinds of goals is easy if you use Matomo. 

  Just head to the Goals section of the dashboard, click Manage Goals and then click the green Add A New Goal button. 

  Fill in the screen below, and add a Goal Revenue at the bottom of the page. Doing so will mean Matomo can automatically calculate the value of each touchpoint when using your attribution model. 

  

  If your analytics platform allows it, make sure you also set up Event Tracking, which will allow you to analyse how many users start to take a desired action (like filling in a form) but never complete the task. 

  Try Matomo for Free

  Get the web insights you need, without compromising data accuracy.

  Start for free

  No credit card required

  Test and validate

  As we’ve explained, linear attribution is a great model in some scenarios, but it can fall short if you have a long or complex sales funnel. Even if you’re sure it’s the right model for your company, testing and validating is important. 

  Ideally, your chosen attribution tool should make this process pretty straightforward. For example, Matomo’s Marketing Attribution feature makes comparing and contrasting three different attribution models easy. 

  Here we compare the performance of three attribution models—linear, first-touch, and last-non-direct—in Matomo’s Marketing Attribution dashboard, providing straightforward analysis.

  

  If you think linear attribution accurately reflects the value of your channels, you can start to analyse the insights it generates. If not, then consider using another attribution model.

  Don’t forget to take action from your marketing efforts, either. Linear attribution helps you spot the channels that contribute most to conversions, so allocate more resources to those channels and see if you can improve your conversion rate or boost your ROI. 

  Make the most of marketing attribution with Matomo

  A linear attribution model lets you measure every touchpoint in your customer journey. It’s an easy attribution model to start with and lets you identify and optimise your most effective marketing channels. 

  However, accurate data is essential if you want to benefit the most from marketing attribution data. If your web analytics solution doesn’t play nicely with cookies or uses sampled data, then your linear model isn’t going to tell you the whole story. 

  That’s why over 1 million sites trust Matomo’s privacy-focused web analytics, ensuring accurate data for a comprehensive understanding of customer journeys.

  Now you know what linear attribution modelling is, start employing the model today by signing up for a free 21-day trial, no credit card required. 

  Try Matomo for Free

  21 day free trial. No credit card required.

  
  

  Your email address
  

  Your website address
  

  Your Analytics subdomain will be .matomo.cloud (change)

  Choose your analytics subdomain
  

  .matomo.cloud

  I accept the terms and conditions and data processing agreement (DPA)

  
  Your information will be used to create an account on our cloud service. For more information please consult our privacy policy.
  
  
  

  Start improving your websites now

  

• Attribution Tracking (What It Is and How It Works)

  23 février 2024, par Erin

  Facebook, TikTok, Google, email, display ads — which one is best to grow your business ? There’s one proven way to figure it out : attribution tracking.

  Marketing attribution allows you to see which channels are producing the best results for your marketing campaigns.

  In this guide, we’ll show you what attribution tracking is, why it’s important and how you can leverage it to accelerate your marketing success.

  What is attribution tracking ?

  By 2026, the global digital marketing industry is projected to reach $786.2 billion.

  With nearly three-quarters of a trillion U.S. dollars being poured into digital marketing every year, there’s no doubt it dominates traditional marketing.

  The question is, though, how do you know which digital channels to use ?

  By measuring your marketing efforts with attribution tracking.

  

  So, what is attribution tracking ?

  Attribution tracking is where you use software to keep track of different channels and campaign efforts to determine which channel you should attribute conversion to.

  In other words, you can (and should) use attribution tracking to analyse which channels are pushing the needle and which ones aren’t.

  By tracking your marketing efforts, you’ll be able to accurately measure the scale of impact each of your channels, campaigns and touchpoints have on a customer’s purchasing decision.

  If you don’t track your attribution, you’ll end up blindly pouring time, money, and effort into activities that may or may not be helpful.

  Attribution tracking simply gives you insight into what you’re doing right as a marketer — and what you’re doing wrong.

  By understanding which efforts and channels are driving conversions and revenue, you’ll be able to properly allocate resources toward winning channels to double down on growth.

  Matomo lets you track attribution across various channels. Whether you’re looking to track your conversions through organic, referral websites, campaigns, direct traffic, or social media, you can see all your conversions in one place.

  Try Matomo for Free

  Get the web insights you need, without compromising data accuracy.

  Start for free

  No credit card required

  Why attribution tracking is important

  Attribution tracking is crucial to succeed with your marketing since it shows you your most valuable channels.

  It takes the guesswork out of your efforts.

  You don’t need to scratch your head wondering what made your campaigns a success (or a failure).

  While most tools show you last click attribution by default, using attribution tracking, or marketing attribution, you can track revenue and conversions for each touchpoint.

  For example, a Facebook ad might have no led to a conversion immediately. But, maybe the visitor returned to your website two weeks later through your email campaign. Attribution tracking will give credit over longer periods of time to see the bigger picture of how your marketing channels are impacting your overall performance.

  Here are five reasons you need to be using attribution tracking in your business today :

  

  1. Measure channel performance

  The most obvious way attribution tracking helps is to show you how well each channel performs.

  When you’re using a variety of marketing channels to reach your audience, you have to know what’s actually doing well (and what’s not).

  This means having clarity on the performance of your :

  • Emails
  • Google Ads
  • Facebook Ads
  • Social media marketing
  • Search engine optimisation (SEO)
  • And more

  Attribution tracking allows you to measure each channel’s ROI and identify how much each channel impacted your campaigns.

  It gives you a more accurate picture of the performance of each channel and each campaign.

  With it, you can easily break down your channels by how much they drove sales, conversions, signups, or other actions.

  With this information, you can then understand where to further allocate your resources to fuel growth.

  2. See campaign performance over longer periods of time

  When you start tracking your channel performance with attribution tracking, you’ll gain new insights into how well your channels and campaigns are performing.

  The best part — you don’t just get to see recent performance.

  You get to track your campaign results over weeks or months.

  For example, if someone found you through Google by searching a question that your blog had an answer to, but they didn’t convert, your traditional tracking strategy would discount SEO.

  But, if that same person clicked a TikTok ad you placed three weeks later, came back, and converted — SEO would receive some attribution on the conversion.

  Using an attribution tracking tool like Matomo can help paint a holistic view of how your marketing is really doing from channel to channel over the long run.

  Try Matomo for Free

  Get the web insights you need, without compromising data accuracy.

  Start for free

  No credit card required

  3. Increase revenue

  Attribution tracking has one incredible benefit for marketers : optimised marketing spend.

  When you begin looking at how well your campaigns and your channels are performing, you’ll start to see what’s working.

  Attribution tracking gives you clarity into the performance of campaigns since it’s not just looking at the first time someone clicks through to your site. It’s looking at every touchpoint a customer made along the way to a conversion.

  By understanding what channels are most effective, you can pour more resources like time, money and labour into those effective channels.

  By doubling down on the winning channels, you’ll be able to grow like never before.

  Rather than trying to “diversify” your marketing efforts, lean into what’s working.

  This is one of the key strategies of an effective marketer to maximise your campaign returns and experience long-term success in terms of revenue.

  4. Improve profit margins

  The final benefit to attribution tracking is simple : you’ll earn more profit.

  Think about it this way : let’s say you’re putting 50% of your marketing spend into Facebook ads and 50% of your spend into email marketing.

  You do this for one year, allocating $500,000 to Facebook and $500,000 to email.

  Then, you start tracking attribution.

  You find that your Facebook ads are generating $900,000 in revenue. 

  That’s a 1,800% return on your investment.

  Not bad, right ?

  Well, after tracking your attribution, you see what your email revenue is.

  In the past year, you generated $1.7 million in email revenue.

  That’s a 3,400% return on your investment (close to the average return of email marketing across all industries).

  In this scenario, you can see that you’re getting nearly twice as much of a return on your marketing spend with email.

  So, the following year, you decide to go for a 75/25 split.

  Instead of putting $500,000 into both email and Facebook ads and email, you put $750,000 into email and $250,000 into Facebook ads.

  You’re still diversifying, but you’re doubling down on what’s working best.

  The result is that you’ll be able to get more revenue by investing the same amount of money, leaving you with higher profit margins.

  Different types of marketing attribution tracking

  There are several types of attribution tracking models in marketing.

  Depending on your goals, your business and your preferred method, there are a variety of types of attribution tracking you can use.

  Here are the six main types of attribution tracking :

  

  1. Last interaction

  Last interaction attribution model is also called “last touch.”

  It’s one of the most common types of attribution. The way it works is to give 100% of the credit to the final channel a customer interacted with before they converted into a customer.

  This could be through a paid ad, direct traffic, or organic search.

  One potential drawback of last interaction is that it doesn’t factor in other channels that may have assisted in the conversion. However, this model can work really well depending on the business.

  2. First interaction

  This is the opposite of the previous model.

  First interaction, or “first touch,” is all about the first interaction a customer has with your brand.

  It gives 100% of the credit to the channel (i.e. a link clicked from a social media post). And it doesn’t report or attribute anything else to another channel that someone may have interacted with in your marketing mix.

  For example, it won’t attribute the conversion or revenue if the visitor then clicked on an Instagram ad and converted. All credit would be given to the first touch which in this case would be the social media post. 

  The first interaction is a good model to use at the top of your funnel to help establish which channels are bringing leads in from outside your audience.

  3. Last non-direct

  Another model is called the last non-direct attribution model. 

  This model seeks to exclude direct traffic and assigns 100% credit for a conversion to the final channel a customer interacted with before becoming a customer, excluding clicks from direct traffic.

  For instance, if someone first comes to your website from an emai campaignl, and then, a week later, directly visits and buys a product, the email campaign gets all the credit for the sale.

  This attribution model tells a bit more about the whole sales process, shedding some more light on what other channels may have influenced the purchase decision.

  4. Linear

  Another common attribution model is linear.

  This model distributes completely equal credit across every single touchpoint (that’s tracked). 

  Imagine someone comes to your website in different ways : first, they find it through a Google search, then they click a link in an email from your campaign the next day, followed by visiting from a Facebook post a few days later, and finally, a week later, they come from a TikTok ad. 

  Here’s how the attribution is divided among these sources :

  • 25% Organic
  • 25% Email
  • 25% Facebook
  • 25% TikTok ad

  This attirubtion model provides a balanced perspective on the contribution of various sources to a user’s journey on your website.

  5. Position-based

  Position-based attribution is when you give 40% credit to both the first and last touchpoints and 20% credit is spread between the touchpoints in between.

  This model is preferred if you want to identify the initial touchpoint that kickstarted a conversion journey and the final touchpoint that sealed the deal.

  The downside is that you don’t gain much insight into the middle of the customer journey, which can make it hard to make effective decisions.

  For example, someone may have been interacting with your email newsletter for seven weeks, which allowed them to be nurtured and build a relationship with you.

  But that relationship and trust-building effort will be overlooked by the blog post that brought them in and the social media ad that eventually converted them.

  6. Time decay

  The final attribution model is called time decay attribution.

  This is all about giving credit based on the timing of the interactions someone had with your brand.

  For example, the touchpoints that just preceded the sale get the highest score, while the first touchpoints get the lowest score.

  For example, let’s use that scenario from above with the linear model :

  • 25% SEO
  • 25% Email
  • 25% Facebook ad
  • 25% Organic TikTok

  But, instead of splitting credit by 25% to each channel, you weigh the ones closer to the sale with more credit.

  Instead, time decay may look at these same channels like this :

  • 5% SEO (6 weeks ago)
  • 20% Email (3 weeks ago)
  • 30% Facebook ad (1 week ago)
  • 45% Organic TikTok (2 days ago)

  One downside is that it underestimates brand awareness campaigns. And, if you have longer sales cycles, it also isn’t the most accurate, as mid-stage nurturing and relationship building are underlooked. 

  Leverage Matomo : A marketing attribution tool

  Attribution tracking is a crucial part of leading an effective marketing strategy.

  But it’s impossible to do this without the right tools.

  A marketing attribution tool can give you insights into your best-performing channels automatically. 

  

  One of the best marketing attribution tools available is Matomo, a web analytics tool that helps you understand what’s going on with your website and different channels in one easy-to-use dashboard.

  With Matomo, you get marketing attribution as a plug-in or within Matomo On-Premise or for free in Matomo Cloud.

  The best part is it’s all done with crystal-clear data. Matomo gives you 100% accurate data since it doesn’t use data sampling on any plans like Google Analytics.

  To start tracking attribution today, try Matomo’s 21-day free trial. No credit card required.

  Try Matomo for Free

  21 day free trial. No credit card required.

  
  

  Your email address
  

  Your website address
  

  Your Analytics subdomain will be .matomo.cloud (change)

  Choose your analytics subdomain
  

  .matomo.cloud

  I accept the terms and conditions and data processing agreement (DPA)

  
  Your information will be used to create an account on our cloud service. For more information please consult our privacy policy.
  
  
  

  Start improving your websites now

  

• When I use ffmpeg to go from a video to frames, and then back to video, the duration is different between the videos

  24 février 2024, par bluepanda

  I am trying to use ffmpeg to convert from a .mp4 (or .mov) video into individual frames, do some processing on those frames, and then convert back to .mp4. The problem is that the resulting video I create is a different duration than the input - I can see this visually when I play the two videos side by side. The difference is not large (i.e. 00:00:00.50 for the input video and 00:00:00.52 for the output video), but when the videos are looped next to each other they get out of sync.

  


  Here is information about the input video retrieved using fluent-ffmpeg's ffmpeg.ffprobe(videoPath) :

  


  metadata {
  streams: [
    {
      index: 0,
      codec_name: 'h264',
      codec_long_name: 'H.264 / AVC / MPEG-4 AVC / MPEG-4 part 10',
      profile: 'High',
      codec_type: 'video',
      codec_tag_string: 'avc1',
      codec_tag: '0x31637661',
      width: 1080,
      height: 1920,
      coded_width: 1080,
      coded_height: 1920,
      closed_captions: 0,
      has_b_frames: 2,
      sample_aspect_ratio: 'N/A',
      display_aspect_ratio: 'N/A',
      pix_fmt: 'yuv420p',
      level: 40,
      color_range: 'tv',
      color_space: 'bt709',
      color_transfer: 'bt709',
      color_primaries: 'bt709',
      chroma_location: 'left',
      field_order: 'unknown',
      refs: 1,
      is_avc: 'true',
      nal_length_size: 4,
      id: 'N/A',
      r_frame_rate: '30000/1001',
      avg_frame_rate: '27000/1001',
      time_base: '1/30000',
      start_pts: 0,
      start_time: 0,
      duration_ts: 15100,
      duration: 0.503333,
      bit_rate: 5660223,
      max_bit_rate: 'N/A',
      bits_per_raw_sample: 8,
      nb_frames: 36,
      nb_read_frames: 'N/A',
      nb_read_packets: 'N/A',
      tags: [Object],
      disposition: [Object]
    },
    {
      index: 1,
      codec_name: 'aac',
      codec_long_name: 'AAC (Advanced Audio Coding)',
      profile: 'LC',
      codec_type: 'audio',
      codec_tag_string: 'mp4a',
      codec_tag: '0x6134706d',
      sample_fmt: 'fltp',
      sample_rate: 48000,
      channels: 2,
      channel_layout: 'stereo',
      bits_per_sample: 0,
      id: 'N/A',
      r_frame_rate: '0/0',
      avg_frame_rate: '0/0',
      time_base: '1/48000',
      start_pts: 0,
      start_time: 0,
      duration_ts: 24160,
      duration: 0.503333,
      bit_rate: 248416,
      max_bit_rate: 'N/A',
      bits_per_raw_sample: 'N/A',
      nb_frames: 27,
      nb_read_frames: 'N/A',
      nb_read_packets: 'N/A',
      tags: [Object],
      disposition: [Object]
    }
  ],
  format: {
    filename: '/Users/name/images/input.mp4',
    nb_streams: 2,
    nb_programs: 0,
    format_name: 'mov,mp4,m4a,3gp,3g2,mj2',
    format_long_name: 'QuickTime / MOV',
    start_time: 0,
    duration: 0.503333,
    size: 963879,
    bit_rate: 15319941,
    probe_score: 100,
    tags: {
      major_brand: 'mp42',
      minor_version: '1',
      compatible_brands: 'isommp41mp42',
      creation_time: '2024-02-14T01:21:12.000000Z'
    }
  },
  chapters: []
}


  


  and here is from running ffprobe directly :

  


  ffprobe '/Users/name/images/input.mp4'
ffprobe version 6.1.1 Copyright (c) 2007-2023 the FFmpeg developers
  built with Apple clang version 15.0.0 (clang-1500.1.0.2.5)
  configuration: --prefix=/opt/homebrew/Cellar/ffmpeg/6.1.1_2 --enable-shared --enable-pthreads --enable-version3 --cc=clang --host-cflags= --host-ldflags='-Wl,-ld_classic' --enable-ffplay --enable-gnutls --enable-gpl --enable-libaom --enable-libaribb24 --enable-libbluray --enable-libdav1d --enable-libharfbuzz --enable-libjxl --enable-libmp3lame --enable-libopus --enable-librav1e --enable-librist --enable-librubberband --enable-libsnappy --enable-libsrt --enable-libssh --enable-libsvtav1 --enable-libtesseract --enable-libtheora --enable-libvidstab --enable-libvmaf --enable-libvorbis --enable-libvpx --enable-libwebp --enable-libx264 --enable-libx265 --enable-libxml2 --enable-libxvid --enable-lzma --enable-libfontconfig --enable-libfreetype --enable-frei0r --enable-libass --enable-libopencore-amrnb --enable-libopencore-amrwb --enable-libopenjpeg --enable-libopenvino --enable-libspeex --enable-libsoxr --enable-libzmq --enable-libzimg --disable-libjack --disable-indev=jack --enable-videotoolbox --enable-audiotoolbox --enable-neon
  libavutil      58. 29.100 / 58. 29.100
  libavcodec     60. 31.102 / 60. 31.102
  libavformat    60. 16.100 / 60. 16.100
  libavdevice    60.  3.100 / 60.  3.100
  libavfilter     9. 12.100 /  9. 12.100
  libswscale      7.  5.100 /  7.  5.100
  libswresample   4. 12.100 /  4. 12.100
  libpostproc    57.  3.100 / 57.  3.100
Input #0, mov,mp4,m4a,3gp,3g2,mj2, from '/Users/name/images/input.mp4':
  Metadata:
    major_brand     : mp42
    minor_version   : 1
    compatible_brands: isommp41mp42
    creation_time   : 2024-02-14T01:21:12.000000Z
  Duration: 00:00:00.50, start: 0.000000, bitrate: 15319 kb/s
  Stream #0:0[0x1](und): Video: h264 (High) (avc1 / 0x31637661), yuv420p(tv, bt709, progressive), 1080x1920, 5660 kb/s, 26.97 fps, 29.97 tbr, 30k tbn (default)
    Metadata:
      creation_time   : 2024-02-14T01:21:12.000000Z
      handler_name    : Core Media Video
      vendor_id       : [0][0][0][0]
      encoder         : AVC Coding
  Stream #0:1[0x2](und): Audio: aac (LC) (mp4a / 0x6134706D), 48000 Hz, stereo, fltp, 248 kb/s (default)
    Metadata:
      creation_time   : 2024-02-14T01:21:12.000000Z
      handler_name    : Core Media Audio
      vendor_id       : [0][0][0][0]


  


  And this is my command to go from video to frames :

  


  ffmpeg -i /Users/name/images/input.mp4 -y -f image2 /Users/name/images/frames/%d.png


  


  After which I convert the frames back to video with this - note that I get by seeing avg_frame_rate is 27000/1001 = 26.97302697 :

  


  ffmpeg -r 26.973026973026972 -i /Users/name/images/frames/%d.png -y -r 26.973026973026972 -b:v 5660223k -f mp4 -pix_fmt yuv420p -t 0.503333 /Users/name/images/output.mp4


  


  And if I then run fluent-ffmpeg's ffmpeg.ffprobe(videoPath) I get :

  


  metadata {
  streams: [
    {
      index: 0,
      codec_name: 'h264',
      codec_long_name: 'H.264 / AVC / MPEG-4 AVC / MPEG-4 part 10',
      profile: 'High',
      codec_type: 'video',
      codec_tag_string: 'avc1',
      codec_tag: '0x31637661',
      width: 1080,
      height: 1920,
      coded_width: 1080,
      coded_height: 1920,
      closed_captions: 0,
      has_b_frames: 2,
      sample_aspect_ratio: '1:1',
      display_aspect_ratio: '9:16',
      pix_fmt: 'yuv420p',
      level: 62,
      color_range: 'unknown',
      color_space: 'unknown',
      color_transfer: 'unknown',
      color_primaries: 'unknown',
      chroma_location: 'left',
      field_order: 'unknown',
      refs: 1,
      is_avc: 'true',
      nal_length_size: 4,
      id: 'N/A',
      r_frame_rate: '27000/1001',
      avg_frame_rate: '27000/1001',
      time_base: '1/27000',
      start_pts: 0,
      start_time: 0,
      duration_ts: 14014,
      duration: 0.519037,
      bit_rate: 52138429,
      max_bit_rate: 'N/A',
      bits_per_raw_sample: 8,
      nb_frames: 14,
      nb_read_frames: 'N/A',
      nb_read_packets: 'N/A',
      tags: [Object],
      disposition: [Object]
    }
  ],
  format: {
    filename: '/Users/name/images/output.mp4',
    nb_streams: 1,
    nb_programs: 0,
    format_name: 'mov,mp4,m4a,3gp,3g2,mj2',
    format_long_name: 'QuickTime / MOV',
    start_time: 0,
    duration: 0.52,
    size: 3383708,
    bit_rate: 52057046,
    probe_score: 100,
    tags: {
      major_brand: 'isom',
      minor_version: '512',
      compatible_brands: 'isomiso2avc1mp41',
      encoder: 'Lavf60.3.100'
    }
  },
  chapters: []
}


  


  and here is from running ffprobe directly :

  


  ffprobe '/Users/name/images/output.mp4'
ffprobe version 6.1.1 Copyright (c) 2007-2023 the FFmpeg developers
  built with Apple clang version 15.0.0 (clang-1500.1.0.2.5)
  configuration: --prefix=/opt/homebrew/Cellar/ffmpeg/6.1.1_2 --enable-shared --enable-pthreads --enable-version3 --cc=clang --host-cflags= --host-ldflags='-Wl,-ld_classic' --enable-ffplay --enable-gnutls --enable-gpl --enable-libaom --enable-libaribb24 --enable-libbluray --enable-libdav1d --enable-libharfbuzz --enable-libjxl --enable-libmp3lame --enable-libopus --enable-librav1e --enable-librist --enable-librubberband --enable-libsnappy --enable-libsrt --enable-libssh --enable-libsvtav1 --enable-libtesseract --enable-libtheora --enable-libvidstab --enable-libvmaf --enable-libvorbis --enable-libvpx --enable-libwebp --enable-libx264 --enable-libx265 --enable-libxml2 --enable-libxvid --enable-lzma --enable-libfontconfig --enable-libfreetype --enable-frei0r --enable-libass --enable-libopencore-amrnb --enable-libopencore-amrwb --enable-libopenjpeg --enable-libopenvino --enable-libspeex --enable-libsoxr --enable-libzmq --enable-libzimg --disable-libjack --disable-indev=jack --enable-videotoolbox --enable-audiotoolbox --enable-neon
  libavutil      58. 29.100 / 58. 29.100
  libavcodec     60. 31.102 / 60. 31.102
  libavformat    60. 16.100 / 60. 16.100
  libavdevice    60.  3.100 / 60.  3.100
  libavfilter     9. 12.100 /  9. 12.100
  libswscale      7.  5.100 /  7.  5.100
  libswresample   4. 12.100 /  4. 12.100
  libpostproc    57.  3.100 / 57.  3.100
Input #0, mov,mp4,m4a,3gp,3g2,mj2, from '/Users/name/images/output.mp4':
  Metadata:
    major_brand     : isom
    minor_version   : 512
    compatible_brands: isomiso2avc1mp41
    encoder         : Lavf60.3.100
  Duration: 00:00:00.52, start: 0.000000, bitrate: 52153 kb/s
  Stream #0:0[0x1](und): Video: h264 (High) (avc1 / 0x31637661), yuv420p(progressive), 1080x1920 [SAR 1:1 DAR 9:16], 52138 kb/s, 26.97 fps, 26.97 tbr, 27k tbn (default)
    Metadata:
      handler_name    : VideoHandler
      vendor_id       : [0][0][0][0]
      encoder         : Lavc60.3.100 libx264


  


  This seems like it should be a fairly common scenario, but I have not been able to find examples of this, and the other questions about incorrect durations on Stack Overflow are about bigger differences (i.e. 3 seconds instead of 10 seconds : Wrong video duration when recording with ffmpeg).

  


  Some other details :

  


  

  • I am running this through a Node.js script with fluent-ffmpeg, but I have also tried running the commands directly in the terminal and the result is the same.

  


  • I am fine with the output frames being .png / .jpg / other formats.

  


  • I am fine with setting this to a different frame rate than the original as long as the two output videos end up with the same duration.

  


  • One suspicious thing is that I set -t 0.503333 when creating the video, but it doesn't seem to work as the result video shows duration: 0.519037 / 00:00:00.52.

  


  


  Thank you for any help !